Ecoenzymatic stoichiometry as a temporally integrated indicator of nutrient availability in soils

ABSTRACT

The extent to which soil enzyme activity in assessing soil nutrient availability is useful has been controversial. In this review, we discuss the utility of ecoenzymatic stoichiometry (i.e. the ratio of nutrient- to carbon (C)-acquiring enzyme activities) on the basis of the resource allocation model for ecoenzyme synthesis. Both the selection of appropriate enzymes and the balance between relative amounts of substrates and enzymes in the soil are decisive factors in utilizing the ecoenzymatic stoichiometry. Ecoenzymatic stoichiometry can evaluate the availability of nitrogen (N), phosphorus, and sulfur in many soils in which the enzyme catalytic reactions are substrate-limited but not enzyme-limited. However, the ecoenzymatic stoichiometry approach does not seem to be applicable in soils where microbes are limited by factors other than nutrient availability, such as low temperature, where the enzyme catalytic reactions are enzyme-limited. Certain enzymes, such as N-acetyl-β-glucosaminidase and protease, appear to be insensitive to soil N availability, because they release compounds containing both N and C which serve as sources for both N and C/energy. We propose the use of enzymes such as L-asparaginase and urease as N-acquiring enzymes that release a compound containing N but not C (i.e. NH₄⁺) as the hydrolysis product. Ecoenzymatic stoichiometry can be considered as an indicator of long-term (weeks) temporally integrated soil nutrient availability, rather than instantaneous availability, for plants as well as microbes, because of (i) the long-term persistence of extracellular enzymes in soils; (ii) a significant correlation between ecoenzymatic stoichiometry and the measurements reflecting the quantity of long-term available nutrients in soil; and (iii) a significant correlation between ecoenzymatic stoichiometry and plant nutrient uptake. This review also identifies challenges in assessing microbial nutrient limitation using ecoenzymatic stoichiometry. With a comprehensive understanding of underlying mechanisms and limitations, ecoenzymatic stoichiometry can be used as a biologically relevant indicator of nutrient availability in combination with other approaches such as conventional chemical extraction methods and the nutrient addition approach.

KEY WORDS:

• Ecoenzymatic stoichiometry
• nutrient availability
• nutrient limitation
• soil enzyme
• vector analysis

1. Introduction

Nitrogen (N), phosphorus (P), and sulfur (S) are essential nutrients for primary production in terrestrial ecosystems. However, excessive application of N and P fertilizer in arable soils has led to water eutrophication (Carpenter 2005). Also, there is a growing concern about the exhaustion of high-quality P rock (Cordell and White 2014). To make fertilizer use more efficient and reduce its loading on the environment, accurate evaluation of nutrient availability in the soil is indispensable. Furthermore, global climate change will affect nutrient cycles in terrestrial environments (Xu et al. 2024). Therefore, it is essential to evaluate nutrient availability for a comprehensive understanding of the mechanisms governing nutrient cycling in both natural and arable soils.

Various tests have been designed to estimate the availability of N (Griffin 2008) and P (Maguire, Chardon, and Simard 2005) in soils, but such soil tests have not been standardized (in this article, nutrient availability to microbes and to plants are synonymous as discussed later). For example, various extractants have been proposed to provide an estimate of available P in soils (Beegle 2005; Hedley 2008). However, the principle for assessing P availability varies among soil P tests and, therefore, inconsistent results have often been obtained between these tests (e.g., Kato, Zapata, and Fardeau 1995; Otani and Ae 1996; Sharpley, Sims, and Pierzynski 1994; Vimpany et al. 1997). Also, the appropriate test may be different among soil types. For example, it was found that P availability estimated using either Truog or modified Bray 2 tests should be treated differently even within Andosols (allophanic vs. non-allophanic) (Ito, Kikawa, and Saigusa 2011). In addition, nutrient cycling is largely mediated by soil microbes. Hence, the complementary use of a biologically relevant indicator with a conventional soil test may provide a more precise evaluation of nutrient availability in soils. Recently, some microbiological properties have been inferred to serve as biological indicators of soil fertility (Hermans et al. 2017; Lemanski, Armbruster, and Bonkowski 2019; Suzuki et al. 2005). Soil enzyme activities also have the potential to indicate nutrient availability in soils, because microbes rely on enzymes to degrade complex substrates into usable forms to take up and because microbial enzyme synthesis is controlled by nutrient availability.

Under a broad definition, ‘soil enzymes’ consist of both intracellular and extracellular enzymes in soils because, at present, clear experimental distinction between intracellular and extracellular enzyme activities in soils is difficult (Gianfreda and Ruggiero 2006; Nannipieri et al. 2011). Under a narrow definition, soil enzymes include (i) abiontic enzymes, which occur in free form dissolved in water and stabilized forms adsorbed to the organic matter and mineral colloids in soil (Chróst 1991) and which are no longer associated with viable cells (Nannipieri, Trasar-Cepeda, and Dick 2018), and (ii) ectoenzymes, which are cell-surface-bound or periplasmic enzymes (Chróst 1991). ‘Ecoenzyme’ is defined as encompassing all enzymes located outside the cell membranes (Sinsabaugh and Follstad Shah 2012); the definition is analogous to that of soil enzymes under the narrow definition. We use the terms ‘Ecoenzyme’ to indicate extracellular enzymes, including ectoenzymes, and ‘soil enzyme’ to represent both intracellular and extracellular enzymes in soils. Microbes are the major source of soil enzymes (Dick and Burns 2011). In the soil environment, activities of free ecoenzymes constitute only a minor fraction of total soil enzyme activities (Buckley et al. 2019), but both activities are strongly correlated with each other (Levakov et al. 2021). Free ecoenzymes are short-lived but can reach organic matter in sub-micron pore spaces, which microbes cannot access (Quiquampoix and Burns 2007), whereas stabilized and persistent ecoenzymes, which adsorb to soil particles, may have a role in detecting substrate input (Wallenstein and Burns 2011). In soil enzyme measurement, activity indicates potential activity, rather than in situ activity, and corresponds to the enzyme concentration in soil, because the enzyme activity is measured under conditions where there is a sufficient concentration of its substrate (Wallenstein and Weintraub 2008). Hence, measured enzyme activity provides a comparative measure of microbial resources directed toward acquiring a nutrient (Sinsabaugh 2005).

Despite the potential of soil enzyme activity to indicate nutrient availability in soils, questions have been raised regarding whether soil enzyme activity really is a sensitive indicator of nutrient availability in soils because the results obtained have been controversial (e.g., Fatemi et al. 2016; Nannipieri, Trasar-Cepeda, and Dick 2018). In this review, we examine the usefulness of soil enzyme activity and particularly ecoenzymatic stoichiometry (or soil enzymatic stoichiometry) as an indicator of soil nutrient availability and microbial nutrient demand, and we also clarify the uncertainties in its use.

2. Soil enzyme activity alone is not adequate for evaluating nutrient availability in soils

There is a contradiction to be resolved in using soil enzyme activity as an indicator of nutrient availability in soils: higher soil enzyme activity can be simultaneously interpreted as evidence of greater nutrient availability or as evidence of nutrient limitation (Fierer, Wood, and de Mesquita 2021; Margenot and Wade 2023). Higher soil enzyme activity may result in higher nutrient availability owing to the greater decomposition of the substrate. For example, it was reported that the activity of N-acquiring enzymes corresponded to the NH₄⁺ production rate in boreal forest and tundra soils (Hofmockel et al. 2010). However, higher enzyme activity also can be induced by low nutrient availability. In ‘microbial economics,’ the benefits of enzyme synthesis and secretion should outweigh the costs (Allison and Vitousek 2005; Allison et al. 2011) because the production of ecoenzymes is a costly energetic process and their secretion means the loss of the carbon (C) and nutrients contained in the enzymes from the microbes (Allison et al. 2011; Sistla and Schimel 2012; Zuccarini et al. 2023). Approximately 1%–5% of C and N assimilation in microbial communities was estimated to be used to make enzymes for secretion into the environment (Allison et al. 2011; Sinsabaugh and Follstad Shah 2012). This can lead to the strict regulation of ecoenzyme synthesis that should depend on the soil nutrient availability in microbes: the lack of a certain nutrient in its available form would induce the synthesis of the nutrient-acquiring ecoenzyme, whereas high nutrient availability would decrease the synthesis. In fact, for N-acquiring enzymes, Sinsabaugh (2005) suggested that the production is repressed by high availability of NH₄⁺ and glutamine, while N deprivation induces the production in microbes. Also, relative P deficiency caused by C and N addition to soil elicited remarkable increases in the activities of two P-acquiring enzymes, alkaline phosphomonoesterase (ALP) and acid phosphomonoesterase (ACP), while the increase in β-D-glucosidase (BG) activity was smaller than that of the P-acquiring enzymes (Figure 1(a)). Nakas et al. (1987) also suggested that the increased microbial population, caused by the addition of C and N, synthesized additional inducible phosphatase to obtain the P they needed from less available sources. However, conflicting results have been reported for P-acquiring enzymes, for example, positive, negative, or no relationships between enzyme activity and P concentration (Jones and Oburger 2011). Hence, the usefulness of soil enzyme activity as an indicator of nutrient availability remains questionable (Fatemi et al. 2016; Fierer, Wood, and de Mesquita 2021; Trasar-Cepeda, Leirós, and Gil-Sotres 2008).

Figure 1. Effects of the addition of carbon as glucose and nitrogen as NH₄Cl on (a) alkaline phosphomonoesterase (ALP), acid phosphomonoesterase (ACP), and β-D-glucosidase (BG) activities and (b) ratios of ALP/BG and ACP/BG activities in an arable Andosol (Mise et al. 2020). Error bars are standard errors. ** = significant at p < 0.01 and *** = significant at p < 0.001 in the Student’s t-test.

There are two possible explanations for these conflicting results. First, a relative balance between the contents of substrates and enzymes largely influences the correspondence between enzyme activity and nutrient availability in soils. When a large amount of substrates is present relative to that of the enzyme degrading it, the reaction rate increases in proportion to the enzyme amount in the soil, potentially yielding a positive correlation between the enzyme activity and the reaction product, e.g., nutrient availability when the enzyme catalyzes the final step in the mineralization process. In contrast, when only a small amount of substrate is present relative to that of the enzyme in the soil, which corresponds to low nutrient availability, microbes would produce a large amount of the enzyme to acquire the nutrient, resulting in a negative correlation between the enzyme activity and the nutrient availability. So far, several studies have reported that enzyme catalytic reactions were substrate-limited rather than enzyme-limited for most of the soils. For example, hydrolysis of organic P (Po) was reported to be limited by substrate availability, not by enzyme availability in various soils (Gressel and McColl 1997; Jarosch et al. 2019). For protein degradation, the reaction was limited by the substrate availability, not by the content of protease (PR) in different soils (Brzostek and Finzi 2012; Greenfield et al. 2020; Hofmockel et al. 2010; Noll et al. 2019; Tateno 1988). Also, for cellulase, the amount of substrate, not the amount of enzymes, limited the reaction in forest soils (Tateno 1988). The exceptions include (i) soils with a large input of nutrient-rich plant litter and (ii) soils (e.g., tundra soil) in which microbial activity is limited by factors other than nutrient availability such as low temperature and high moisture content. In the former case (i), for example, in soils with a pulse of P-rich plant litter, the synthesis of phosphatases would rise owing to increased microbial growth and P demand (Figure 2(a)). Therefore, the reaction product – inorganic P (Pi) – would also increase because the reaction is enzyme-limited, not substrate-limited, and thus it initially culminates in a positive relationship between phosphatase activity and P availability (Geisseler and Horwath 2009). However, the positive relationship between phosphatase activity and P availability could gradually change to a negative relationship as the plant litter decomposes, because a great amount of phosphatase caused by the input of P-rich plant litter would degrade the substrate (Po compounds), and then the resulting lower availability of Po (that is, a change from an enzyme-limited condition to a substrate-limited condition) would decrease the amount of the reaction product, Pi (Figure 2(a)). The opposite trend is expected in soils when a large amount of P-poor plant litter enters the soil because the relative P availability should be lowered due to the P deficiency in microbes (Figure 2(b)), like a P-deficient condition caused by C and N addition with resultant high activities of ALP and ACP (Figure 1(a)). In the latter case (ii), such as tundra soils with a large content of organic matter (Melle et al. 2015), because microbes are limited by low temperature and high moisture content, the enzyme catalytic reactions might be enzyme-limited, not substrate-limited, which might lead to a positive relationship between enzyme activity and reaction product, e.g., nutrient availability. Hofmockel et al. (2010) reported that the NH₄⁺ production rate was controlled by the enzyme content, not by substrate availability in tundra and boreal forest soils.

Figure 2. Schematic diagram for the effects of inputs of (a) P-rich plant litter and (b) P-poor plant litter on P-acquiring enzyme activity and its hydrolysis product level (i.e., inorganic P; Pi) in soil. Arrows indicate the time point of litter input. P is present mainly as organic P in plant litter. Input of P-rich litter results in a positive relationship between P-acquiring enzyme activity and Pi concentration, whereas input of P-poor litter leads to a negative relationship. See the text for details.

Second, microbial biomass in soils should affect the relationship between enzyme activity and nutrient availability. In the case of using soil samples with a large difference in microbial biomass, no negative correlation between the enzyme activity and the nutrient availability might be observed, even if microbes produced a considerable amount of enzyme to acquire a deficient nutrient because, in general, enzyme production is proportional to biomass. Soil enzyme activity can be high in soils with large microbial biomass even if high nutrient availability lowers the enzyme synthesis per unit biomass, and vice versa. In this case, a positive correlation should be observed between soil enzyme activity and nutrient availability in soils. As an extreme example, when comparing a soil with low organic matter and low microbial biomass in which low P availability raises the synthesis of phosphatase per unit microbial biomass to a soil with high organic matter and high microbial biomass in which high P availability lowers the enzyme synthesis per unit microbial biomass, phosphatase activity on a soil weight basis may be lower in the former soil (Figure 3). Accordingly, soil enzyme activity alone is not adequate for evaluating nutrient availability and microbial nutrient demand in soils.

Figure 3. Schematic diagram depicting the comparison between a soil with low active microbial biomass in which low P availability increases the synthesis of phosphatase per unit biomass and a soil with high active microbial biomass in which high P availability lowers the phosphatase synthesis per unit biomass. See the text for details.

To adjust for the effects of differences in microbial biomass on soil enzyme activity, the ratio of enzyme activity to microbial biomass (i.e., microbial biomass-specific enzyme activity) has been sometimes employed (e.g., Dove et al. 2020; Zhou, Wang, and Jin 2017). However, microbial biomass-specific enzyme activity might not be a good indicator of nutrient availability in soil, because it can depend not only on microbial nutrient demand but also on the proportion of active microbial biomass to total microbial biomass. The difference in the proportion of active microbial biomass between soils prohibits a direct comparison of the microbial biomass-specific activity as an indicator of nutrient availability between soils. It is well known that most soil microbes are dormant (Blagodatskaya and Kuzyakov 2013; Lennon and Jones 2011), probably because of the oligotrophic condition in most soil environments even with high organic matter (Hobbie and Hobbie 2013; Morita 1988; Tateno 1988).

3. Evaluation of microbial nutrient demand and soil nutrient availability on the basis of the resource allocation model

Sinsabaugh and Moorhead (1994) proposed the resource allocation model for ecoenzyme synthesis, in which microbes will allocate their resources to C-, N-, and P-acquiring enzyme syntheses in response to the soil nutrient status to maximize their productivity. The model assumes that many taxa use similar regulatory pathways to tie ecoenzyme synthesis to nutrient availability (Sinsabaugh 2005). The resource allocation model suggests that relative investment in the synthesis of each nutrient-acquiring enzyme indicates microbial nutrient demand and thus provides an insight into the nutrient availability in the environment. For example, when N availability is low, microbes invest more resources into N-acquiring enzyme synthesis, which leads to a higher ratio of N- to C-acquiring enzyme activities, whereas P-acquiring enzymes are produced more in comparison with other nutrient-acquiring enzymes when P availability is low, resulting in a larger ratio of P- to C-enzyme activities (Figure 4). Indeed, a relative P deficiency caused by C and N addition to the soil leads to high ratios of ALP/BG and ACP/BG activities (Figure 1(b)), suggesting that microbes invested more resources in the synthesis of P-acquiring enzymes than in C-acquiring enzymes in response to the relatively lower availability of P. Note that this stoichiometric approach avoids the influence of microbial biomass on enzyme activity in the soil.

Figure 4. Relative resource allocation in the synthesis of C-, N-, and P-acquiring enzymes under low N availability or low P availability. Modified from Fujita et al. (2019b).

The ecoenzymatic stoichiometry (i.e., the ratio of nutrient- to C-acquiring enzyme activities) on the basis of the resource allocation model has been widely used to evaluate microbial nutrient demand and nutrient availability in soils (Sinsabaugh and Follstad Shah 2012). Despite this, the enzymatic ratio does not always accurately predict nutrient availability in soils (Fierer, Wood, and de Mesquita 2021). This can be because an inadequate enzyme was employed, especially for N-acquiring enzymes for some studies, as discussed in Section 4. For example, if microbes synthesize certain enzymes to acquire not only nutrients but also C/energy, then the ratio of the nutrient- to C-acquiring enzyme activities may not be applicable as an indicator of the nutrient availability. According to the simulation results of Averill and Classen (2014), the ratio of N- to C-acquiring enzyme activities decreases with increasing N availability in the ‘EnzOpt model,’ which assumes that certain N-acquiring enzymes are produced to acquire only N. When N availability is high, the enzymatic ratio remains constant in the ‘EnzMax model,’ which assumes that it is produced to acquire both N and C (Figure 5). Thus, it is essential to select the appropriate enzyme that fits the EnzOpt model to predict soil nutrient availability. In addition, the allocation model should not be applied to soils with enzyme limitation rather than substrate limitation, as described in Section 2. Further, inappropriate soil tests have often been used to estimate nutrient availability in soils (e.g., using a test suitable for alkaline soils on acidic soils), which may conceal the association between ecoenzymatic stoichiometry and true nutrient availability in soils. So far, few studies have been conducted on which soil nutrient forms and/or nutrient availability index microbes respond to when investing their resources in the nutrient-acquiring enzyme synthesis. We discuss the validity of the resource allocation model for ecoenzyme synthesis across different soil types and land uses. Also, we discuss which nutrient form microbes respond to when investing their resources in ecoenzyme synthesis in the soils.

Figure 5. Conceptual diagram showing the relationship between ecoenzymatic stoichiometry and nutrient availability, on the basis of the simulation result of Averill and Classen (2014). The case of an N-acquiring enzyme is illustrated. EnzOpt model: the enzyme synthesis is optimized to match the microbial N demand. EnzMax model: C acquisition is prioritized from substrates when N availability is high.

3.1. P-acquiring enzymes

3.1.1. ALP

Two phosphatases, ALP and ACP, have been extensively studied as P-acquiring enzymes, with both releasing PO₄³⁻ from phosphomonoesters (Acosta-Martínez and Tabatabai 2011). It is widely known that ALP dominates in alkaline soils, whereas ACP prevails in acid soils (Denison 2000; Nannipieri et al. 2011). Yet, ALP might play a pivotal role even in acid soils; ALP activity was reported to be comparable with or even higher than ACP activity in some arable acid soils (Mise et al. 2018; Moro, Kunito, and Sato 2015), probably owing to lime application. Also, metagenome analysis revealed that the abundance of ALP genes was higher than that of ACP genes in an acid forest soil of pH 3.8 (Bergkemper et al. 2016).

Moro et al. (2015) and Fujita et al. (2017) reported that the ratio of ALP/BG activities was negatively and most strongly correlated with Truog-P concentration (Figure 6(c)) among various soil P tests in arable soils. This negative correlation was consistent with the resource allocation model (Sinsabaugh and Follstad Shah 2012; Sinsabaugh and Moorhead 1994), in which microbes preferentially expend resources to synthesize P-acquiring enzymes when soil P availability is low. The Truog test assesses available Pi (Maguire, Chardon, and Simard 2005), thereby inferring that the ratio of ALP/BG activities should increase as Pi availability decreases in the arable soils. It should be mentioned that a useful soil P test should differ between soils, depending on soil types and land use (see also Section 5); a dilute acid extraction procedure such as the Truog test cannot be used for evaluating P availability in alkaline or calcareous soils (Fixen and Grove 1990; Kuo 1996) or in acid forest soils, as described later.

Figure 6. Relationships between (a) the ratio of L-asparaginase (LA) to β-D-glucosidase (BG) activities and potentially mineralizable N concentration in arable and forest soils (Fujita et al. 2018); (b) the ratio of urease (UR) to BG activities and potentially mineralizable N concentration in arable and forest soils (Fujita et al. 2018); (c) the ratio of alkaline phosphomonoesterase (ALP) to BG activities and Truog-P concentration in arable soils (Fujita et al. 2017; Mise et al. 2018; Moro, Kunito, and Sato 2015; Otsuka and Kunito 2021); (d) the ratio of acid phosphomonoesterase (ACP) to BG activities and H₂O-extractable inorganic P (H₂O-Pi) concentration in acid forest soils (Fujita et al. 2017; Kunito et al. 2012b, 2016); and (e) the ratio of arylsulfatase (ARS) to BG activities and soluble and adsorbed sulfate concentration in arable and forest soils (Kunito et al. 2022). The r and p indicate Spearman’s rank correlation coefficient and its p-value.

In contrast, the ratio of ALP/BG activities did not show a significant correlation with any Po fractions in arable soils (Fujita et al. 2017; Moro, Kunito, and Sato 2015), indicating that microbial synthesis of ALP relative to that of BG would respond to Pi availability rather than Po concentration. Renz et al. (1999) also reported that Po, the substrate for phosphatases, was not the major parameter responsible for the synthesis of the enzymes. These findings were in accord with the mechanism of ALP production. Soil ALP is primarily produced and secreted into the soil by bacteria (Nannipieri et al. 2011) and the bacterial ALPs are classified into three types, PhoA, PhoD, and PhoX. The genes (phoA, phoD, and phoX, respectively) encoding these ALPs are included in Pho regulons, the transcription of which is known to be controlled by Pi concentration (Monds et al. 2006; Vershinina and Znamenskaya 2002; Yuan et al. 2006). According to Park et al. (2022), the extracellular Pi level, which can be sensed by Pi binding proteins such as PstS on the cell surface (Hsieh and Wanner 2010; Qi et al. 2016), is the responsible signal that regulates the Pho regulon. Karl (2014) also reported that Pho regulon phosphatases were induced by Pi limitation rather than by the presence of hydrolyzable Po.

In contrast, no significant correlation between the ALP/BG ratio and Pi concentration was observed in acid forest soils (Fujita et al. 2017). This might be because of the low ALP synthesis in forest soils with low pH, because it is likely that ALP with high pH optimum would not function under such acidic conditions (Fujita et al. 2017). In addition, the small amount of synthesis of ALP by fungi (Nannipieri et al. 2011), which dominate in acid forest soils (Binkley and Fisher 2013), might also contribute to the observation.

It should be noted that, in the relationship shown in Figure 6(c), the ALP/BG ratio is compared with the soil P concentration, not with the ratio of soil P to C concentrations. At present, no accurate method for estimating soil C availability exists. DOC concentration has often been used as an index of C availability, but its usefulness can be questioned (see Section 7-2). Because soil microbes are usually C (energy)-limited even when the organic matter pool appears large (Morita 1988; Sinsabaugh 2005; Smith and Paul 1990; Tate 2000), C-acquiring enzymes would be preferentially synthesized compared with other nutrient-acquiring enzymes independent of the soil C level. However, even with C limitation, P should also be needed for soil microbes, and they would invest the resources in the synthesis of P-acquiring enzymes in response to the availability of Pi. In this case, the ALP/BG ratio might depend more on P concentration than on ratio of P to C concentrations, which might bring about the significant negative correlation between the ALP/BG ratio and Pi concentration (Figure 6(c)). Such a view might also be applicable to other nutrient-acquiring enzymes shown in Figure 6 and is the subject of further investigation.

It is notable that the relationship between ALP activity and P availability in soils tends to be weaker than those of other nutrient-acquiring enzymes (p < 0.1; Figure 6) (see also discussion for ACP). Mori (2024) and Mori et al. (2024) pointed out that ratio of BG/phosphomonoesterase activities cannot be used as a universal indicator of soil P availability. There are two plausible explanations for the weaker correlation. First, the presence of constitutively produced ALP weakens the association between ALP activity and P availability in soils (Duhamel et al. 2021). According to Lidbury et al (2021, 2022), many bacteria in the phylum Bacteroidetes produce constitutive ALP (PafA) at high levels. Second, ALP might be synthesized to acquire C as well as P in microbes (Chróst 1991), which would translate into a weaker response of ALP synthesis to P availability (McConnell, Kaye, and Kemanian 2020). When Po compounds are used as C/energy sources, polar PO₄ groups may hinder direct Po uptake (McConnell, Kaye, and Kemanian 2020), although some Po compounds such as glucose-6-phosphate and fructose-6-phosphate can be directly taken up by microbes only if they are present at high concentrations (Park et al. 2022). Consequently, phosphatase-mediated dephosphorylation of Po is needed to use these Po compounds as C/energy sources (McConnell, Kaye, and Kemanian 2020), which is independent of P availability. It was also reported that transcription of Pho regulon phosphatases was affected not only by Pi levels but also by C availability in some microbes (Choi and Saier 2005; Santos-Beneit 2015). Indeed, a large amount of ALP was synthesized to satisfy the C/energy requirements through dephosphorylation of the targeted Po compounds before its microbial uptake in the deep sea (Colman et al. 2005; Karl 2014). Thus, ALP synthesis may be partly due to C/energy acquisition, but the relationship between ALP activity and P availability concurred with the EnzOpt model, but not with the EnzMax model (Figure 5). The result suggested that ALP might be synthesized primarily to acquire P and to a lesser extent to acquire C.

3.1.2. ACP

The ACP/BG ratio showed a significant negative correlation with Pi availability in acid forest soils (Figure 6(d)). The ACP/BG ratio did not exhibit a significant correlation with any Po fractions in forest soils (Fujita et al. 2017), which concurred with the results of the ALP/BG ratio as mentioned above. The finding implied that, in acid forest soils, the microbes preferentially expended resources in the ACP synthesis relative to the BG synthesis when Pi availability was low. Many of the ACPs that are synthesized by fungi which are dominant in forest soils (Binkley and Fisher 2013) belong to phytases, a group of phosphomonoesterase. For example, most of the 13 genes encoding extracellular ACP in Aspergillus oryzae were classified as a class of phytases – histidine acid phosphatase (Marui et al. 2012). Many phytases are included in Pho regulons, the transcription of which is known to be controlled by Pi concentration in fungi (Greiner 2007; Hill and Richardson 2007; Yip et al. 2023). However, in arable soils, no significant correlation was found between the ACP/BG ratio and P availability (Fujita et al. 2017). Spohn and Kuzyakov (2013) reported that the response of ACP producers to P availability was smaller than that of the ALP producers, and that P fertilization reduced ALP activity but did not change ACP activity in arable soils. In contrast, P fertilization reduced ACP activity and thus the ACP/BG ratio in tropical rain forest soils (Turner and Wright 2014). In addition, a higher bacteria to fungi ratio in arable compared with forest soils (Weil and Brady 2017) might also contribute to the lack of a significant association between the ACP/BG ratio and P availability in the arable soils. It is because some bacterial ACPs such as class A, a dominant bacterial ACP (nonspecific acid phosphohydrolases) in soils (Lidbury et al. 2017; Neal et al. 2018), and class B are known to be Pi-irrepressible (Rossolini et al. 1998).

A significant negative relationship between the ACP/BG ratio and Pi availability has also been observed in pasture soils (Bajouco et al. 2020) and waterlogged soils (Fujita et al. 2020; Kunito et al. 2018). Furthermore, the significant negative association was also discerned in arable soils if the soil properties were similar between soil samples (Moro, Kunito, and Sato 2015). Further research is needed to clarify which land use the resource allocation model is applicable to in relation to ACP synthesis.

There are two possible explanations for the correlation being not as robust between the ACP/BG ratio and P availability (Figure 6(d)): (i) some ACPs are constitutively synthesized (Rossolini et al. 1998), and (ii) ACPs are synthesized to acquire not only P but also C to some extent (McConnell, Kaye, and Kemanian 2020), similar to the discussion for ALP.

3.2. N-acquiring enzymes

Several studies have suggested that microbial synthesis of N-acquiring enzymes (N-acetyl-β-glucosaminidase (NAG) and leucine aminopeptidase (LAP) are usually measured as N-acquiring enzymes) seems to have a weaker relationship with N availability in comparison with P-acquiring enzyme synthesis (Sinsabaugh and Follstad Shah 2012; Sullivan et al. 2014; Zechmeister-Boltenstern et al. 2015), although this depends on the N-acquiring enzyme types (Chen et al. 2018a; Jian et al. 2016). This may be because the mineralization of organic N compounds is often coupled to C/energy acquisition (McGill and Cole 1981; Olander and Vitousek 2000; Sinsabaugh and Follstad Shah 2012; Sullivan et al. 2014), and because, unlike P-acquiring enzymes which have wide substrate affinities, each N compound is decomposed by distinct enzyme systems, as discussed by Sinsabaugh et al. (1993). However, Fujita et al. (2018) revealed that microbial resource allocation for the synthesis of two N-acquiring enzymes – L-asparaginase (LA) (Figure 6(a)) and urease (UR) (Figure 6(b)) – was strongly dependent on N availability across land use (i.e., forest and arable), but not for NAG and PR. These results suggest that microbes synthesize LA and UR in response to soil N availability in arable and forest soils. Microbes have sensors on the cell surface to monitor extracellular levels of NH₄⁺ and amino acids (Bahn et al. 2007; Forsberg and Ljungdahl 2001). The negative relationships of the LA/BG and UR/BG ratios with N availability (Figure 6(a, b)) corresponded well with the EnzOpt model (Averill and Classen 2014) in Figure 5, suggesting that synthesis of LA and UR was optimized to acquire N, but not C, in the soils. Unlike the LA/BG and UR/BG ratios, the PR/BG and NAG/BG ratios did not show significant negative correlations with available N concentration (Fujita et al. 2018). Hence, synthesis of PR and NAG might not be controlled primarily by N availability in the soils.

This difference in response to soil N availability between N-acquiring enzymes can be explained by their hydrolysis products. LA catalyzes the hydrolysis of L-asparagine and carboxyl-terminal asparagine peptides, releasing ammonia (Dunlop, Meyer, and Roon 1980; Howard and Carpenter 1972). UR hydrolyzes urea – a fertilizer and a well-known product of general purine catabolism – to yield ammonia (Mobley and Hausinger 1989). In contrast, the hydrolysis products of PR and NAG (peptides and amino acids, and N-acetylglucosamine, respectively) contain not only N but also C (Beier and Bertilsson 2013; Landi et al. 2011). Thus, LA and UR are considered to acquire only N, but PR and NAG seem to acquire C as well as N, as also discussed in Section 4. This assumption is supported by the underlying mechanism for the synthesis of these enzymes in microbes. It has been reported that LA is synthesized to acquire N but not to acquire C in Bacillus licheniformis (Golden and Bernlohr 1985). Atkinson and Fisher (1991) and Fisher and Sonenshein (1991) reported that syntheses of both LA and UR were increased when the NH₄⁺ level was low but were decreased by high levels of NH₄⁺ and amino acids in Bacillus subtilis. Synthesis of extracellular LA was increased by N starvation, whereas intracellular LA was constitutively produced in bacteria and fungi (Dunlop and Roon 1975; Fisher and Wray 2002). Thus, high activity at low N availability might be attributed to the extracellular LA, and low activity at high N availability might be because of intracellular LA in the soils (Figure 6(a)). McCarty et al. (1992) reported the suppression of microbial UR production when adding inorganic N or amino acids to soils. According to Mobley and Hausinger (1989), microbial regulation of UR production is classified into three types: in Type 1 the synthesis is activated under N-limiting conditions, whereas the synthesis is repressed in the presence of NH₄⁺ or N-rich compounds including urea; in Type 2 the synthesis is induced by the presence of the substrate (i.e., urea); and in Type 3 the enzyme is synthesized constitutively. The negative relationship between UR/BG and N availability shown in Figure 6(b) might imply that Type 1 UR-producers were predominant in the soils. In contrast to the case of LA and UR, microbial synthesis of extracellular PR was induced not only by N limitation but also by C or S limitation in bacteria (Allison and Macfarlane 1990; Whooley, O’Callaghan, and McLoughlin 1983) and fungi (North 1982; Wiame, Grenson, and Arst 1985). Also, production of NAG was under C catabolite repression in bacteria (Keyhani and Roseman 1999) and a fungus Penicillium chrysogenum (Pócsi et al. 1993).

According to Fujita et al. (2018, 2019a), the ratio of N-acquiring enzyme to P-acquiring enzyme activities was related to the available N/P ratio in the soils. The results suggest that microbes allocate their resources to N- and P-acquiring enzyme synthesis in response to relative N and P availabilities in the soils.

3.3. S-acquiring enzymes

Arylsulfatase (ARS) is one of the S-acquiring enzymes produced by many bacteria and fungi (Kertesz, Fellows, and Schmalenberger 2007; Santana et al. 2021; Slezack-Deschaumes et al. 2012), and its natural substrates are sulfate esters, including aromatic, aliphatic, and carbohydrate sulfates, from which SO₄²⁻ is released by hydrolysis (Kertesz 1999). Consistent with the resource allocation model (Sinsabaugh and Follstad Shah 2012; Sinsabaugh and Moorhead 1994), the ARS/BG ratio exhibited a significant negative correlation with soluble and adsorbed SO₄²⁻ concentration across land use (i.e., forest and arable) (Figure 6(e)), indicating that more resources were invested to produce ARS in comparison with BG synthesis when S availability was low, and vice versa. A similar relationship was observed in tropical forest soils (Wang et al. 2023a). The mechanism underlying the negative link was that ARS synthesis was induced under conditions of S limitation (Dodgson, White, and Fitzgerald 1982; Gahan and Schmalenberger 2014; Santana et al. 2021) and was repressed by excess SO₄²⁻ (Kertesz 1999; Santana et al. 2021; Stressler et al. 2016) in many bacteria and fungi.

Organic S has been assumed to be mineralized either via biochemical mineralization of sulfate esters – substrates for ARS – to satisfy the microbial S demand or by biological mineralization of C-bonded S to gain C and energy, with SO₄²⁻ being released as a byproduct (McGill and Cole 1981). The inverse relationship between the ARS/BG ratio and S availability (Figure 6(e)) was highly consistent with the EnzOpt model (Figure 5), suggesting that ARS synthesis was optimized to acquire SO₄²⁻, but not C, in the soils. Scherer (2009) also suggested that activities of sulfatases were controlled by the S demand of microbes rather than their need for energy.

It is notable that the resource allocation model for N-, P-, and S-enzymes was applicable to rewetted soils that were stored air-dried (Fujita et al. 2017, 2018, 2020; Kunito et al. 2022) as well as to moist soils (e.g., Fujita, Miyabara, and Kunito 2019a; Kunito et al. 2018; Moro, Kunito, and Sato 2015). It is also noteworthy that the resource allocation model is not applicable for the enzyme activity ratio when the enzyme reactions are limited by enzyme availability rather than by substrate availability, as described in Section 2.

4. Appropriate enzymes for the resource allocation model

To assess the microbial nutrient demand on a community scale, the enzymes used for ecoenzymatic stoichiometric analyses need to be produced by many microbes. For C-acquiring enzymes, BG was harbored by more than 80% of sequenced bacterial lineages (Berlemont and Martiny 2013), and most of the fungi also had cellulolytic enzymes including BG (Berlemont 2017). Wang et al. (2023b) found an average of three gene copies of BG per bacterial cell in metagenomic analysis of 32 soils. For P-acquiring enzymes, approximately half of bacteria had phoD, one of the genes encoding ALP (Lidbury et al. 2017; Neal et al. 2017), and approximately 20%–50% of bacteria harbored the ACP class A gene in soils (Lidbury et al. 2017). Many fungi also possess phosphatase genes (Treseder and Lennon 2015). For N-, and S-acquiring enzymes, many bacteria and fungi produce LA (Arima et al. 1972), UR (Mobley and Hausinger 1989; Mobley, Island, and Hausinger 1995), and ARS (Kertesz, Fellows, and Schmalenberger 2007; Santana et al. 2021; Slezack-Deschaumes et al. 2012).

Until now, NAG and LAP have been widely used as N-acquiring enzymes in the resource allocation model (e.g., Hill et al. 2012; Moorhead et al. 2013; Sinsabaugh et al. 2009, 2008). However, the use of NAG and LAP as simply N-acquiring enzymes may be problematic. Sinsabaugh et al. (2012) suggested that NAG and LAP were synthesized by microbes to acquire both N and C, because their hydrolysis products (i.e., amino sugar and amino acids, respectively) contain not only N but also C. A similar point has been made by other researchers (e.g., Bárta et al. 2014; Camenzind et al. 2020; Castle et al. 2017; Fujita et al. 2018; Margenot and Wade 2023; Mori 2022; Schleuss et al. 2019; Stone, Plante, and Casper 2013). Mori et al. (2018) reported that NAG and LAP synthesis were likely driven by C acquisition, rather than N acquisition in N-fertilized tropical rainforest soils. The hydrolysis products of these enzymes were likely to be used as C sources when N was abundant but used as N sources when N was the limiting factor (Averill and Classen 2014; Geisseler, Joergensen, and Ludwig 2012; Yang et al. 2016). It was also inferred that NAG acts as a C-acquiring enzyme when fungal chitin and bacterial peptidoglycan are increased with the progress of plant litter decomposition (Mori 2020; Mori et al. 2021). The same point is applicable to PR: it was reported that PR was produced to acquire C rather than N in soils (Geisseler and Horwath 2008; Norman et al. 2020; Vranova, Rejsek, and Formanek 2013). The inappropriate use of these enzymes as simply N-acquiring enzymes appears to introduce errors that may compromise the prediction of microbial nutrient demand and nutrient availability on the basis of ecoenzymatic stoichiometry (Fierer, Wood, and de Mesquita 2021; Moorhead et al. 2023). Therefore, reconsidering and refining the enzymes included in ecoenzymatic stoichiometric analyses are necessary (Weintraub 2023). We recommend the use of enzymes such as LA and UR as N-acquiring enzymes, because their hydrolysis product is a compound containing N but not C (i.e., NH₄⁺).

So far, BG has been generally used as a C-acquiring enzyme in ecoenzymatic stoichiometric analyses (Hill et al. 2012; Moorhead et al. 2013; Sinsabaugh et al. 2009, 2008). BG is involved in the hydrolysis of cellobiose, which is the main product in hydrolysis of cellulose by cellulases (Tabatabai 1994). The capacity for producing BG is broadly distributed among bacteria and fungi as mentioned above. The presence of substrate is critical for enzyme synthesis (Geisseler and Horwath 2008, 2009; Wallenstein and Burns 2011). Polysaccharides – cellulose and hemicellulose – comprise the pivotal C/energy sources for microbes in most of the soils: more than 20% of soil organic C (SOC) was plant-derived, even in soils with little SOC, where microbially derived C was greater than plant-derived C (Huang et al. 2023). Chen et al. (2018b) reported that the cellulose derived from crop straw added to soil was still present after 12 months. Most of the BG are also known to be relatively nonspecific and involved in hydrolysis not only of cellulose but also of hemicellulose (Baldrian and Valášková 2008; Sørensen et al. 2013). According to Yang et al. (2019), BG activity was linked more to microbial biomass than to cellulose content. Furthermore, glucose usually does not accumulate at high enough levels to cause the catabolite repression of BG in soils (Aro, Pakula, and Penttilä 2005). Considering these characteristics, BG seems to be a useful indicator of microbial resource allocation to acquire C (Sinsabaugh et al. 2012). Mori et al. (2023) suggested that using a single BG as a C-acquiring enzyme might be inappropriate to sufficiently represent the complex cascade of enzyme depolymerization. However, Moorhead et al. (2013) and Arnosti et al. (2014) suggested that the suite of the related enzymes that hydrolyze cellulose tends to be correlated with each other, and a single enzyme (e.g., BG) has emerged as a proxy for the actions of many. This view is supported by the underlying molecular mechanisms for producing carbohydrate-active enzymes (CAZymes) including BG. Many CAZymes consist of one operon in bacteria (Gardner and Schreier 2021) and many CAZymes are under the control of the same transcription system in fungi (Rytioja et al. 2014).

A similar discussion is applicable to other nutrient-acquiring enzymes. In most studies, the enzymes that are typically measured represent a small subset of potentially important enzymes (Fierer, Wood, and de Mesquita 2021). In these cases, a single enzyme may not be taken as a proxy of entire metabolic processes involving many different enzymes in soils (Mori, Rosinger, and Margenot 2023; Nannipieri, Trasar-Cepeda, and Dick 2018). For example, several of the enzymes involved in protein degradation are not closely related to each other (Nannipieri et al. 2012). Margenot and Wade (2023) also noted that most enzymes do not represent a terminal step that releases a plant-available form of an element and thus do not reflect nutrient status. Therefore, caution should be applied in considering a single enzyme activity in relation to the nutrient mineralization process. However, in the case of N-acquiring enzymes, for example, if the synthesis of an enzyme is regulated by the soil NH₄⁺ level, measuring that single enzyme (e.g., LA and UR) would enable the evaluation of N availability in soils, although it should be noted that the rate of organic N degradation cannot be evaluated.

The discussion in this section shows that the reaction products and the mechanisms that regulate enzyme synthesis are crucial in selecting appropriate enzymes for ecoenzymatic stoichiometric analyses. Although Moorhead et al. (2023) suggested that ecoenzymatic stoichiometry methods are best suited to natural ecosystems where plant-derived organic matter is the main source of energy and nutrients for microbes but not to croplands with reduced plant litter inputs and added fertilizers, the discussion above shows that the ecoenzymatic stoichiometry can be used to evaluate nutrient availability not only in natural ecosystems but also in arable soils.

5. Ecoenzymatic stoichiometry as a temporally integrated indicator of nutrient availability

Regarding P-acquiring enzymes, the ratio of ALP/BG activities was negatively and most strongly correlated with Truog-P concentration in arable soils (Figure 6(c)) in various soil P tests such as Bray-2P and several fractions in the Hedley sequential extraction procedure (Fujita et al. 2017; Moro, Kunito, and Sato 2015). The Truog test is the most commonly used method for arable soils in Japan. In the Truog test, P associated with the Ca-bearing phase, CaHPO₄, accumulated through the continuous application of lime in arable soils, can be dissolved by the acid solution, but less soluble tricalcium phosphate and hydroxyapatite cannot (Ando et al. 2021; Yamaguchi et al. 2023). Ligand exchange between the aqueous sulfate and the adsorbed phosphate is also involved in the P extraction (Shuai 2018). The easily available Pi estimates using the Truog test are the P form to which microbes respond when investing their resources in P-acquiring enzyme synthesis in arable soils. Because the Truog test is better related to the soil P quantity (available for long term) than to the soil P intensity (currently available) (Fixen and Grove 1990) and was reported to be correlated with P uptake by barley over a 4-week period (Susuki, Lawton, and Doll 1963), the ratio of ALP/BG activities might reflect the long-term (weeks) temporally integrated P availability in arable soils.

In waterlogged soil, the ratio of ACP/BG activities was significantly negatively correlated with the Bray-2P concentration (Fujita et al. 2020; Kunito et al. 2018) but not with the Truog-P concentration (Fujita et al. 2020). The Bray test is known to easily extract Al-associated P because of the high affinity of fluoride used in the Bray test to Al (Fixen and Grove 1990; Kamprath and Watson 1980). In waterlogged soils, P adsorbed on Fe oxides is released primarily as a result of the hydrolysis of Fe(III)–P compounds and the reductive dissolution of Fe oxides onto which P is adsorbed under anaerobic conditions (Kyuma 2004; Sanyal and De Datta 1991) and then re-adsorbed onto redox-stable constituents, such as Al oxides and Al–humus complexes. Hence, the Bray-2P concentration might better represent P availability than Truog-P in anaerobic soils. It was reported that the Bray-2P concentration in paddy soils reflected the uptake of P in rice plants (Uwasawa, Sangtong, and Cholitkul 1988), and it might reflect the soil P quantity that serves as a storage reservoir of readily available P (Fixen and Grove 1990). Thus, the ratio of ACP/BG activities might mirror the temporally integrated P availability in waterlogged soils.

In contrast to the reflection of the soil P quantity through the ratio of ALP/BG activities in arable soils and the ratio of ACP/BG activities in waterlogged soils, the ratio of ACP/BG activities had a significant and negative correlation with H₂O-Pi concentration reflecting soil P intensity (Ballard and Pritchett 1975) in acid forest soils (Figure 6(d)). It is conceivable that weak extractants such as H₂O reflect the intensity of P supply and provide the best index of plant response over short growth periods, whereas the stronger extractants for the Bray and Truog tests are better predictors of the response over longer growth periods in acid soils (Ballard and Pritchett 1975). It is unclear why the ratio of ACP/BG activities reflects P intensity in acid forest soils. We assume that, unlike arable upland and waterlogged soils, acid forest soils without P fertilization might show little temporal change in H₂O-Pi concentration, and thus the ACP synthesis might reflect the most available Pi fraction (i.e., H₂O-Pi) in the forest soils. This deserves further investigation.

The ratio of ARS/BG activities is significantly negatively correlated with soluble and adsorbed SO₄²⁻ concentration across arable and acid forest soils (Figure 6(e)). Barrow (1967) and Jones (1986) reported that soluble and adsorbed SO₄²⁻ usually correlated best with crop yields and S uptake by crops. Thus, the ratio might reflect the temporally integrated S availability in arable and acid forest soils.

Unlike P- and S-acquiring enzymes, N-acquiring enzyme synthesis reflects the concentration of the organic nutrient form, rather than the inorganic form. Both the LA/BG and UR/BG ratios exhibited significant negative correlations with potentially mineralizable N concentration across arable and forest soils (Figure 6(a, b)) but not with extractable NH₄⁺ and NO₃⁻ (Fujita et al. 2018, 2019a). This result may differ from the expectation on the basis of the microbial response, in which microbes can monitor extracellular levels of NH₄⁺ and amino acids (Bahn et al. 2007; Forsberg and Ljungdahl 2001). Extractable NH₄⁺ and NO₃⁻ represent the easily available N in soils, but they also have a rapid turnover (Binkley and Vitousek 1989). Thus, extractable NH₄⁺ and NO₃⁻ might only represent the available concentration at a single point of time and they might be poor indices of the fluxes of these ions (Binkley and Hart 1989). In contrast, the potentially mineralizable N (mineralized N during 28-day aerobic incubation or mineralized N during 7-day anaerobic incubation) might reflect the N supply rate in soils. According to Rastetter and Shaver (1992) and Sullivan et al. (2014), the rate of nutrient supply is more important than the nutrient pool size for plants and microbes, and it does not necessarily correlate with the nutrient pool size. Especially for N, compared with P and S, the supply rate seems to be more important than the concentration at a given time in regulating the availability in soils, probably because the extractable N level may be more temporally variable than P and S levels in soils. Rastetter and Shaver (1992) suggested that the extractable N concentration was less strongly buffered and more temporally variable than P concentration, and thus the replenishment rate was more vital for N availability than that of P in soils. It was observed that KCl-extractable inorganic N diminished by 94%, while Truog-P slightly increased 2 months after planting leaf radish in arable soils without manure amendment (Fujita, Miyabara, and Kunito 2019a). Chapin et al. (2011) and Ros et al. (2011a, 2011b) also suggest that net N mineralization is a reliable indicator of the soil’s ability to release N for plant growth, because net N mineralization ultimately determines how much can be taken up by plants. Considering these together, potentially mineralizable N reflects temporally integrated N availability, and thus, the LA/BG and UR/BG ratios would be good indicators of temporally integrated N availability in arable and acid forest soils.

The long-term persistence of ecoenzymes in soils may allow ecoenzymatic stoichiometry to reflect the long-term (weeks) temporally integrated soil nutrient availability. Schimel et al. (2017) reported that substantial activity remained measurable for some ecoenzymes even after 12 weeks of incubation under continuous fumigation with CHCl₃ vapor to prevent the synthesis of new enzymes. Because the presence of the stabilized soil enzyme pool obscures short-term shifts in microbial enzyme synthesis allocation in response to altered nutrient input (Waring, Weintraub, and Sinsabaugh 2014), ecoenzymatic stoichiometry may be more representative of long-term patterns of microbial resource investments than of short-term responses to changes in nutrient availability (Weintraub 2023). Moro et al. (2015) suggest that ratio of ALP/BG activities may indicate the integrated response of microbes to P status in the soils over a long timescale, because a substantial portion of the soil enzymes would be present as ecoenzymes associated with clays and humic substances and persist long term and because the ratio reflects the soil P quantity rather than the soil P intensity. This view is also applicable to other ecoenzymatic stoichiometry. Recently, Moorhead et al. (2023) also suggested that ecoenzymatic stoichiometry can represent an integrated response of the microbial community to nutrient availability that reflects both current and past conditions, within the lifetime of ecoenzymes. The persistence period might be different between enzyme types and between soils (Renella et al. 2007). Thus, further study is needed to evaluate how long the ecoenzymatic stoichiometry reflects soil nutrient availability.

6. Usefulness of ecoenzymatic stoichiometry as an indicator of plant nutrient uptake

Plants and microbes often access the same N pools, and high microbial N availability also suggests high plant N availability (Sullivan et al. 2014), although N limitation of microbial activity is not necessarily aligned with N limitation of net primary production (Sullivan et al. 2014). Cui et al. (2022a) and Liu et al. (2023) suggested that the ratio of microbial N/P limitation was significantly correlated with the ratio of plant N/P limitation across Chinese forests (Cui et al. 2022a), although microbial nutrient limitation was inappropriately assessed using ecoenzymatic stoichiometry as discussed in Section 7. Lemanski et al. (2019) reported that microbial nutrient limitation assessed by soil respiration after the application of nutrients served as a predictor of sugar beet yield. Microbial nutrient demand could be used for the assessment of both nutrient availability for plants and plant nutrient uptake.

Ecoenzymatic stoichiometry should indicate microbial nutrient demand and nutrient availability for microbes, rather than for plants, because almost all the ecoenzymes originate from microbes (Dick and Burns 2011). However, Fujita et al. (2018) reported that the N uptake of celery had a significant negative relationship with the ratio of LA/BG activities (Figure 7(a)). Moro et al. (2015) found a significant negative relationship between the P content of cabbage and the ratio of ALP/BG activities in arable soils (Figure 7(b)). These results indicate that plant nutrient uptake is higher when nutrient availability is higher for microbes (i.e., the enzymatic ratio is lower) and vice versa. These connections suggest that ecoenzymatic stoichiometry may be a useful indicator of plant nutrient uptake. This is because plant nutrient uptake corresponds to the nutrient quantity rather than intensity in soils (e.g., Ballard and Pritchett 1975; Sattari et al. 2012; Susuki, Lawton, and Doll 1963) and because the ecoenzymatic stoichiometry reflects the long-term (weeks) temporally integrated soil nutrient availability. The correspondence between plant nutrient uptake and microbial ecoenzymatic stoichiometry should be more rigorously tested at several time points during plant growth in further studies.

Figure 7. Relationships (a) between N uptake of celery and ratio of L-asparaginase (LA) to β-D-glucosidase (BG) activities (Fujita et al. 2018) and (b) between P content of cabbage and ratio of alkaline phosphomonoesterase (ALP) to BG activities (Moro, Kunito, and Sato 2015). Spearman’s rank correlation also showed significant statistical results: r = −0.846, p < 0.001 for relationship in (a) and r = −0.567, p < 0.05 for relationship in (b).

7. Ecoenzymatic stoichiometry to assess microbial nutrient limitation

7.1. Ecoenzymatic stoichiometry and vector analysis

Sinsabaugh et al. (2008, 2009) reported that a global mean of C-:N-:P-acquiring enzyme activity ratio was 1:1:1 in soil as well as in wetlands and rivers. On the basis of this ratio, microbial nutrient limitation was estimated (Figure 8). When the ratio of N- to P-acquiring enzymes is greater than 1, microbes are assumed to be N-limited, whereas when the ratio of C- to N- acquiring enzymes is greater than 1, microbes are thought to be C/energy-limited (Figure 8(a); Hill et al. 2012). For instance, when the C-:N-:P-acquiring enzyme activity ratio is 6:7:4 in a soil, N limitation is expected for the microbial community, as plotted in Figure 8(a).

Figure 8. Estimation of microbial nutrient limitation using (a) ecoenzymatic stoichiometry (Hill et al. 2012) and (b) vector analysis (Moorhead et al. 2013, 2016). Results for the C-:N-:P-acquiring enzyme activity ratio of 6:7:4 are shown as an example, indicating N limitation in both analyses. See the text for details.

To analyze the three enzymes (i.e., C-:N-:P-acquiring enzymes) simultaneously, vector analysis was developed by Moorhead et al. (2013) (Figure 8(b)). The horizontal axis represents the ratio of C-acquiring enzyme to the sum of P- and C-acquiring enzyme activities, and the vertical axis corresponds to ratio of C-acquiring enzyme to the sum of N- and C-acquiring enzyme activities (Figure 8(b)). The relative enzyme activities are expressed as proportions to eliminate undefined values resulting from a zero in the denominator (Moorhead et al. 2016). In this analysis, vector length indicates the relative deficiency of C/energy to other nutrients: a greater vector length indicates a greater C/energy deficiency. Also, a vector angle greater than 45° indicates P limitation, while a vector angle less than 45° exhibits N limitation. As an example, the result for the C-:N-:P-acquiring enzyme activity ratio of 6:7:4 is shown in Figure 8(b). The vector angle is 37.6°, indicating N limitation in the microbial community, while the vector length is 0.757.

Many existing studies have evaluated microbial nutrient limitation on the basis of a general congruence in the relative pattern of C-, N- and P-acquiring enzyme activities as 1:1:1 across terrestrial ecosystems. However, the activities of individual enzymes exhibit considerable variation (Sinsabaugh et al. 2008, 2009). Mori (2022) and Mori et al. (2023) questioned the premise that the 1:1:1 ratio of C-, N- and P-acquiring enzyme activities was the threshold for microbial nutrient limitation. It must be noted that a global mean of C-:N-:P-enzyme ratio as 1:1:1 was obtained using BG, both NAG and LAP, and phosphatase as C-, N-, and P-acquiring enzymes, respectively, from the enzyme measurements using 4-methyl-umbelliferone (MUF)-based substrates. Thus, caution should be taken because this approach is subject to inappropriate use of NAG and LAP as simply N-acquiring enzymes as described in Section 4. Also, it is uncertain whether the enzyme ratio as 1:1:1 is valid for measurements using p-nitrophenol-based substrates, although the relationship was reported to be strong between the measurements by fluorescent microplate assay using MUF-based substrates and by the colorimetric p-nitrophenol bench-scale method (Dick et al. 2018).

7.2. Threshold elemental ratio (TER) approach

The TER is a stoichiometric elemental ratio in the available substrates at which the control of metabolism switches from C/energy supply to nutrient supply (e.g., P) (Sterner and Elser 2002). In this approach, TER for C:N and C:P (TER_C:N and TER_C:P, respectively) is compared with the soil available C:N and C:P (R_C:N and R_C:P, respectively) to assess microbial nutrient limitation. The TER_C:N and TER_C:P are calculated using the following equations (Qiu et al. 2021; Sinsabaugh, Hill, and Follstad Shah 2009; Zheng et al. 2020):

(1) $\mathit{TE}{R}_{\mathit{C}:\mathit{N}}=\mathit{\hspace{0.17em}}(\mathit{EE}{A}_{\mathit{C}:\mathit{N}}\times B_{\mathit{C}:\mathit{N}})/n_0$(1)

(2) $\mathit{TE}{R}_{\mathit{C}:\mathit{P}}=\mathit{\hspace{0.17em}}(\mathit{EE}{A}_{\mathit{C}:\mathit{P}}\times B_{\mathit{C}:\mathit{P}})/p_0$(2)

where EEA_C:N and EEA_C:P are the ratio of C- to N-acquiring enzyme activities and the ratio of C- to P-acquiring enzyme activities, respectively; B_C:N and B_C:P are the microbial biomass C:N and C:P ratios, respectively; and n₀ and p₀ are the e^intercepts (the intercept is determined by the regression relationship of C- vs. N-acquiring enzyme activities and C- vs. P-acquiring enzyme activities, respectively). When R_C:N is smaller than TER_C:N or R_C:P is smaller than TER_C:P, the soil microbial community is not limited by N or P, respectively. When R_C:N is greater than TER_C:N or R_C:P is greater than TER_C:P, the microbial community is N- or P-limited, respectively; higher R_C:N – TER_C:N or R_C:P – TER_C:P values equate to greater N or P limitation, respectively.

Similar to vector analysis, appropriate selection of enzymes is indispensable for the TER approach (see section 4). Furthermore, R_C:N and R_C:P vary substantially depending on which method is used to assess soil nutrient availability, and this consequently affects the values of R_C:N – TER_C:N and R_C:P – TER_C:P. Dissolved organic C (DOC), total dissolved N (TDN), and Olsen-P have often been used in the TER approach (e.g., Cui et al. 2022b; Qiu et al. 2021; Zheng et al. 2020). However, the use of DOC as an indicator of C availability may be ambiguous, because it consists of labile and recalcitrant DOC in various proportions (Bolan et al. 2011) and a large portion of DOC is not easily available (De Troyer et al. 2011; Qualls and Haines 1992). In addition, occluded organic C within aggregates, which is protected from enzymatic degradation, can be extracted as DOC (Basile-Doelsch, Balesdent, and Pellerin 2020). A method for assessing available N has not been standardized (Griffin 2008). Fujita et al. (2018) reported that potentially mineralizable N was the more useful indicator for N availability for microbes than TDN, and the concentration of TDN was approximately 7-fold that of potentially mineralizable N in arable soils. Various extractants have been used to estimate available soil P (Maguire, Chardon, and Simard 2005). The concentration depends on the extractant: the concentration of Bray-2P was approximately 8-fold that of Truog-P in arable Vertisols in Spain (López-Piñeiro and Garcia-Navarro 2001), and the Bray-1P concentration was 4.5-fold that of Truog-P and 3.4-fold that of Olsen-P in acid forest soils in U.S.A. (Ballard and Pritchett 1975). Bundy et al. (2005) noted that Olsen method is more suitable for neutral and alkaline soils than for acid soils in assessing soil available P. It must be emphasized that the soil test value is not an absolute concentration but only correlates with the level of available nutrient in the soil (Beegle 2005). Soil enzyme activity might reflect temporally integrated nutrient availability as discussed in Section 5 and, therefore, the extraction methods for measuring the long-term (weeks) temporally integrated nutrient availability in soil might be adequate for evaluating R_C:N and R_C:P.

7.3. Critiques of using ecoenzymatic stoichiometry to assess microbial nutrient limitation

Cui et al. (2023a) point out that the approach using the general congruence of the ratio of C-, N- and P-acquiring enzyme activities as 1:1:1 lacks quantitative thresholds of microbial nutrient limitation and that there is great uncertainty in using ecoenzymatic stoichiometry-based methods to predict nutrient limitation. Then, they define a critical threshold, TER, for microbial nutrient limitation as follows:

In the case of C and N,

(3) $\mathit{TE}{R}_{\mathit{C}:\mathit{N}}=\mathit{\hspace{0.17em}}\mathit{NUE}/\mathit{CUE}\times B_{\mathit{C}:\mathit{N}}$(3)

where NUE is N use efficiency, and CUE is C use efficiency.

From equations (1) and (3)

(4) $\mathit{NUE}/\mathit{CUE}\mathit{\hspace{0.17em}}=\mathit{\hspace{0.17em}}\mathit{EE}{A}_{\mathit{C}:\mathit{N}}/n_0$(4)

Cui et al. (2023a) assumed that n₀ = 1 (for P, also assumed as 1) and also used maximum C, N, P use efficiencies as CUE^max = 0.6 and NUE^max = PUE^max = 0.9 when neither energy nor nutrients are limiting. The threshold was estimated as EEA_C:N <1.5 for N limitation and EEA_C:P <1.5 for P limitation. Also, they estimated the C limitation as EEA_C:N × EEA_C:P >2.25. Using these critical threshold values, the frequency of primary limitation in the soil microbial community was P > N > C limitation (Cui, Moorhead, et al. 2023a). Only 22% of 847 globally distributed sites showed microbial C limitation (Cui, Peng, et al. 2023b). Their result differed from the general theory that soil microbes are considered to be primarily limited by C (Demoling, Figueroa, and Bååth 2007; Kunito, Tobitani, et al. 2012a; Smith and Paul 1990; Soong et al. 2020). It is widely accepted that C/energy utility is small in soils despite the presence of large amount of organic matter, because of its protection resulting from (i) interactions with minerals to form organo-mineral associations (Hemingway et al. 2019); (ii) the physical separation between organic matter and microbes by the occlusion of organic matter with short-range ordered aluminosilicates (Lenhardt et al. 2023) and within aggregates (Basile-Doelsch, Balesdent, and Pellerin 2020); and (iii) the spatially heterogeneous distribution of both organic matter and microbes, leading to reduced microbial encounter with substrates (Lehmann et al. 2020). For example, even in the Arctic tundra with a large amount of organic matter, microbes were reported to be C limited in some sites and in parts of the active season as well as winter (McMahon and Schimel 2017; Melle et al. 2015). Hence, the approach of Cui et al. (2023a, 2023b) may underestimate the global soil microbial C limitation. As a possible explanation for the underestimation, there is a contradiction in the method of Cui et al. (2023a) for quantifying critical thresholds for microbial C, N, and P limitations. In Equation (4)(4) $\mathit{NUE}/\mathit{CUE}\mathit{\hspace{0.17em}}=\mathit{\hspace{0.17em}}\mathit{EE}{A}_{\mathit{C}:\mathit{N}}/n_0$(4) , when N limitation is stronger, NUE/CUE on the left-hand side would be greater (see Figure 1 in Cui et al. (2023a)), but EEA_C:N/ n₀ on the right-hand side would be smaller because of the relatively large production of N-acquiring enzymes compared with C-acquiring enzymes. Furthermore, according to Schimel et al. (2022), biochemically defined CUE is different from that estimated from the activities of extracellular enzymes, and it may be problematic to apply ecoenzymatic stoichiometry to measure CUE. Thus, it seems premature to define a critical threshold for microbial nutrient limitation on the basis of ecoenzymatic stoichiometry.

According to Rosinger et al. (2019), the nutrient addition experiment showed that respiration, the microbial biomass response, and the microbial growth rate in both grassland and forest soils were all C limited, whereas the ecoenzymatic stoichiometry approach showed that microbes in grassland and forest soils were estimated to be P limited and co-limited by C and P, respectively. The findings imply that the sensitivity of the response to nutrient availability varies between microbial properties, and/or the estimated microbial nutrient limitation differs between approaches used. According to Moorhead et al. (2023), different limiting nutrients between nutrient addition experiments and ecoenzymatic stoichiometry are probably due to the difference in response time: hours–days response in the nutrient addition experiment vs. days–weeks response in ecoenzymatic stoichiometry. Also, it seems that the limited nutrients estimated by ecoenzymatic stoichiometry are not perfectly matched with the definition of limited nutrients in other studies (Morris and Blackwood 2015; Tate 2000), which is described as ‘the nutrient that is in lowest supply relative to organismal needs and will limit growth and activity.’ Mori et al. (2023) recommend the direct measurement of the microbial growth response to nutrient additions to accurately determine the nutrient limitation of soil microbes. In contrast to the nutrient addition approach, ecoenzymatic stoichiometry analysis can provide measures of the 'relative' resource use limitation rather than 'actual' resource use limitation for soil microbes, as pointed out by Jing et al. (2020).

Although the nutrient addition approach can be useful for estimating limited nutrients in soil microbes, it cannot be used as a sensitive indicator of the availability status of a soil nutrient when the microbes are limited by another nutrient. For example, when soil microbes are limited by a nutrient other than P even if the availability of P is very low, the nutrient addition approach would fail to appreciate the low P availability (Moro, Kunito, and Sato 2015). In contrast, ecoenzymatic stoichiometry can be gradually varied depending on the relative availability of the nutrient, and therefore it is sensitive enough to evaluate P status in soil microbes (Moro, Kunito, and Sato 2015). Also, the nutrient addition experiment can provide only an estimate of the limiting nutrient at a given time (Moorhead et al. 2023). Thus, ecoenzymatic stoichiometry not only serves as a complement to the nutrient addition approach in assessing microbial nutrient limitation but also provides a temporally integrated assessment of microbial nutrient demand and nutrient availability in soils, as suggested by Moorhead et al. (2023).

8. Issues to consider when using ecoenzymatic stoichiometry and future challenges

When using ecoenzymatic stoichiometry to assess nutrient availability in soils, we need to pay attention to the effects of pH on the stoichiometry. Zuccarini et al. (2023) reported that soil enzyme activity was influenced not only by nutrient availability but also by soil pH, and that not all enzymes were equally sensitive to pH. In fact, soil pH exhibited significant positive correlations with LAP (Sinsabaugh et al. 2008), LA (Ekenler and Tabatabai 2004; Fujita et al. 2018), UR (Ekenler and Tabatabai 2004) and ALP (Fujita et al. 2017) activities, and negative correlations with NAG (Sinsabaugh et al. 2008) and ACP (Kunito et al. 2016; Sinsabaugh et al. 2008) activities. For BG activity, no significant correlation (Sinsabaugh et al. 2008) and a significant negative correlation (Fujita et al. 2018) with pH were obtained. Furthermore, optimum pH differs between enzyme classes (Turner 2010) and between soils even for the same enzyme class (e.g., ACP) (Kunito, Tobitani, et al. 2012a; Trasar-Cepeda and Gil-Sotres 1988). For example, the optimum pH for ACP was well correlated with soil pH, with the optimum value being approximately one pH unit greater than the soil pH (Kunito, Tobitani, et al. 2012a; Trasar-Cepeda and Gil-Sotres 1988). These findings suggest that soil pH affects ecoenzymatic stoichiometry. Fujita et al. (2018) found that the LA/BG and UR/BG ratios were regulated primarily by N availability, but the influence of soil pH was also significant. A similar result was reported for the ALP/BG ratio in arable soils, with the ratio being affected primarily by P availability followed by soil pH (Fujita et al. 2017). Accordingly, soil pH should be considered in the use of ecoenzymatic stoichiometry as an indicator of nutrient availability, particularly when soil pH ranges widely in samples.

The resource allocation model was originally developed for ecoenzymatic synthesis (Sinsabaugh and Follstad Shah 2012; Sinsabaugh and Moorhead 1994), because ecoenzyme secretion means a loss of C and nutrients contained in the enzymes from microbes (Allison et al. 2011; Sistla and Schimel 2012; Zuccarini et al. 2023). The activities of ACP and ALP are likely derived from enzymes that are present extracellularly because the substrates used in the measurements – p-nitrophenyl phosphate (Lidbury et al. 2021) and MUF-phosphate (Luo et al. 2009) – do not cross the cytoplasmic membrane. In contrast, measurements of LA and UR represent the activities of the enzymes that are present both intracellularly and extracellularly. Various bacteria, filamentous fungi, and yeasts synthesize extracellular LA (Arima et al. 1972), but some microbes produce both intracellular and extracellular LA (Dunlop and Roon 1975). Hence, L-asparagine, the substrate of LA, should be mineralized both intracellularly and extracellularly. Amino acids are known to be mineralized to NH₄⁺ before uptake or be directly incorporated into a cell (Barak et al. 1990; Geisseler et al. 2010), releasing unneeded NH₄⁺ when the degradation of amino acids is driven by the need for energy and C (Myrold and Bottomley 2008). For UR, most of the enzymes appear to be cytoplasmic proteins in both fungi and bacteria (Geisseler et al. 2010; Mobley and Hausinger 1989). It was suggested that urea was taken up directly and hydrolyzed intracellularly in soil microbes, rather than mineralized extracellularly (Nielsen, Bonde, and Sørensen 1998), although a large proportion of the UR activity in soil has been found to be extracellular and associated with soil particles (Geisseler et al. 2010). Qin et al. (2010) used the method of Klose and Tabatabai (1999) with modifications in response to criticisms made by Renella et al. (2002) and estimated that approximately 38%–48% of soil UR activity was extracellular. Also, approximately half of ARS activity was estimated to be intracellular in meadow and woodland soils (Margon and Fornasier 2008). For intracellular enzymes, the C and nutrients contained in the enzymes are not lost from microbes and, therefore, are less costly than secreted extracellular enzymes. However, the investment of a large amount of energy (and also C and N before recycling within the cell) is required for the synthesis of intracellular enzymes (Wang and Kuzyakov 2023). Hence, the resource allocation model can be applied not only to extracellular enzymes but also to intracellular enzymes. In fact, all the LA/BG, UR/BG, and ARS/BG ratios were highly consistent with the resource allocation model (Figure 6). Hence, strictly speaking, it seems that the ‘soil enzymatic stoichiometry,’ not the ‘ecoenzymatic stoichiometry,’ might be important for evaluating microbial nutrient demand and soil nutrient availability. The suitability of an enzyme for the resource allocation model may be due mainly to the regulatory mechanism of the enzyme production, as discussed in Section 4, rather than to whether it is an intracellular or extracellular enzyme.

For the increased microbial enzyme activity when soil nutrient availability is low, two explanations arise from the point of view of microbial community composition, which are not mutually exclusive: (i) microbes that were originally present increase the production of the enzymes, and (ii) the proportion of the enzyme-producing microbes are augmented in the community. An example that supports assumption (i) is that microbes increased the transcription of Pho regulons including the genes of several ALPs (phoA, phoX, and phoD) and an ACP (phoN) when the concentration of Pi was low in the environment (Park et al. 2022; Santos-Beneit 2015; Yuan et al. 2006). To support assumption (ii), changes in microbial functional gene abundance can reflect its encoding extracellular enzyme activity in the soil (Chen and Sinsabaugh 2021; Trivedi et al. 2016). There were positive correlations between ALP activity and phoD gene abundance (Chen et al. 2019a, 2019b; Fraser et al. 2015b; Luo et al. 2017), between ALP activity and phoD and phoX gene abundances (Acuña et al. 2016), and between ACP activity and phoC gene abundance (Wang, Xue, and Jiao 2021a), with negative correlations between P availability and phoD gene abundance (Wei et al. 2019) and between P availability and ACP and ALP gene abundance (Cui et al. 2022b). For increased enzyme-producing microbes in response to low nutrient availability, particular types of enzyme-producing microbes might dominate over the other enzyme-producing microbes. Significant negative correlations between ALP activity and the diversity of the phoD gene were found in soils (Figure 9). Similar results have been reported for phoD gene diversity in other soils (Tan et al. 2013; Fraser et al. 2015a; Chen et al. 2019a, 2019b; Wei et al. 2019, but see; Luo et al. 2019), for glycoside hydrolases gene diversities (Malik et al. 2020), and for the chitinase gene diversity (Beier and Bertilsson 2013; Metcalfe et al. 2002). These observations suggest that low nutrient availability might select microbes with particular genotypes, leading to a lower diversity of the enzyme-encoding genes in soils (Tan et al. 2013). We assume that a clear negative correlation between enzyme activity and gene diversity is likely to be obtained only when using similar types of soils such as samples of the same soil that have been subjected to different fertilizer management practices. Future research should be directed toward better understanding the relationship between enzyme activity and its encoding gene diversity in soils.

Figure 9. Relationships between the Shannon-Wiener diversity index (H’) for phoD-harboring communities and alkaline phosphomonoesterase (ALP) activities in (a) arable soils (Kunito et al. 2019; Moro 2015) and (b) waterlogged soils (Fujita et al. 2020). The r and p indicate Pearson’s correlation coefficient and its p-value.

In contrast to the studies mentioned above, no significant correlations were found between ALP activity and phoD gene abundance (Ragot et al. 2017; Wei et al. 2019) and between UR activity and the abundance of ureC (Wang et al. 2018), the UR-encoding gene, in other studies. Fraser et al. (2015b) reported that ALP activity was not correlated with transcript abundance of phoD in soils. These results might be due, at least in part, to the activity of ecoenzymes that have stabilized on soil colloids, persisted for a long time and, therefore, did not directly correlate with the gene abundance in soils (Burns et al. 2013; Wang et al. 2018).

The link between soil enzymatic stoichiometry and microbial community composition appears to be complicated. Because not all microbes possess genes coding a particular enzyme as described in Section 4, the influences of the microbial community composition on the enzyme activity (and soil enzymatic stoichiometry) might be significant (Wang et al. 2021b). Also, even if the overall microbial community is nutrient-limited, it does not necessarily mean that all species in the community are limited by the same nutrient. Indeed, Rosinger et al. (2019) reported that the limiting nutrient can be different between bacteria and fungi in a given soil. However, Kivlin and Hawkes (2020) suggested that variation in microbial composition is unlikely to impinge on aggregated microbial function directly because there are highly functional redundancies in a soil microbial community. Hence, although the microbial community composition might have a significant effect on soil enzymatic stoichiometry, it seems that nutrient availability and soil pH would show greater effects on soil enzymatic stoichiometry than does microbial community composition (e.g., Fujita et al. 2017, 2018).

Recent evidence has pointed to the possibility that quorum-sensing mechanisms might also influence soil enzymatic stoichiometry. Ecoenzyme synthesis was reported to be controlled by quorum sensing as well as by nutrient availability in many bacteria (Cezairliyan and Ausubel 2017; DeAngelis, Lindow, and Firestone 2008). The effect of quorum sensing on ecoenzyme synthesis was demonstrated in soil (McBride and Strickland 2019) and marine (Hmelo, Mincer, and Van Mooy 2011; Van Mooy et al. 2012) environments. The production of ecoenzymes by quorum-sensing mechanisms appears to be a useful means to gain benefits through cooperation among microbes (Traving et al. 2015). In vitro and simulation experiments found that ecoenzyme synthesis might be advantageous overall if bacterial density is sufficiently high, but of little benefit if their density is low (Pai, Tanouchi, and You 2012). However, conflicting results have been obtained, with the synthesis of some enzymes being increased, whereas some are decreased by the quorum sensing mechanism in the environment (Urvoy et al. 2022). Future studies are required to reveal the significance of quorum sensing for ecoenzyme synthesis and stoichiometry in soils.

9. Concluding remarks

To date, inconsistent results have been observed between ecoenzymatic activity and soil nutrient availability in many studies. Most of the discrepancies between studies can be explained using the resource allocation model for ecoenzymatic synthesis. For all the P-, N-, and S-acquiring enzymes, the ecoenzymatic stoichiometry can be used to evaluate microbial nutrient demand and nutrient availability in soils. However, some enzymes that are generally considered as N-acquiring enzymes, such as NAG and PR, did not show significant correlations with soil N availability. Because NAG and PR release N compounds containing C, which is used as not only N but also a C/energy source by microbes, these enzyme activities seem insensitive to soil N availability. In contrast, the synthesis of N-acquiring enzymes that release NH₄⁺, such as LA and UR, does seem to reflect the soil N availability. Additionally, the effects of soil pH are significant for many enzyme activities, and thus, should be considered when evaluating nutrient availability on the basis of ecoenzymatic stoichiometry. It is worth noting that the ecoenzymatic stoichiometry approach cannot be used to assess nutrient availability in soils with a relatively large amount of substrates compared with the amount of the enzyme degrading it. These include (i) soils with a large input of nutrient-rich plant litter and (ii) soils in which microbial activity is limited by factors other than nutrient availability, such as low temperature. It is also noteworthy that estimating actual microbial nutrient limitation using the ecoenzymatic stoichiometry approach seems to be premature.

Many enzymes are involved in the degradation of complex substrates; thus, use of a single enzyme activity will not evaluate the whole mineralization process. Yet, if the production of a particular enzyme is controlled by soil nutrient availability, the single enzyme could represent the soil nutrient availability in the ecoenzymatic stoichiometry approach without measuring the activities of the many enzymes involved in the whole mineralization process.

The ecoenzymatic stoichiometry approach can be used as a biologically relevant indicator of temporally integrated nutrient availability, rather than the instantaneous availability of nutrients, in soils, and thus it can evaluate the nutrient availability for plants as well as for soil microbes. This approach can complement conventional chemical extraction methods in evaluating nutrient availability in a wide range of soils and management systems. In contrast, nutrient addition experiments can provide a snapshot of microbial nutrient limitation at the time of nutrient addition in soils. Both the ecoenzymatic stoichiometry approach and the nutrient addition approach can be used in tandem to assess the precise status of microbial nutrient demand and nutrient availability in soils. Further studies are warranted to delineate the mechanisms of enzyme synthesis at a community level underlying the resource allocation model, including the effects of microbial community composition and quorum sensing.

Research on soil enzymes is becoming increasingly important because global climate change will affect the nutrient cycling mediated by soil enzymes. A recent meta-analysis of studies assessing the impact of global warming on nutrient limitation in plants and soil microbes revealed that warming changes the C:N:P stoichiometry in plants, soils, and microbes, and exacerbates P deficiency for both plants and microbes (Xu et al. 2024). The ecoenzymatic stoichiometry approach, serving as an indicator of temporally integrated nutrient availability in soils, can provide crucial insights into the mechanisms governing C/energy and nutrient cycling in both natural and arable ecosystems, particularly under environmental changes in the future.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by JSPS KAKENHI Grant Number JP23K11389.