The stoichiometry of soil microbial biomass determines metabolic quotient of nitrogen mineralization

We retrieved peer-reviewed articles that simultaneously measured the rate of N mineralization and soil microbial biomass in terrestrial ecosystems by the Web of Science (http://apps.webofknowledge.com) and China National Knowledge Infrastructure Database (http://cnki.net) till 30 July, 2018. We also retrieved articles using Google Scholar. The terms for searching articles were '(nitrogen mineralization OR N mineralization) AND (microbial biomass)'. Duplicates of article were removed. The criteria for selecting eligible articles include: 1. the net N mineralization and MBN were simultaneously reported (approximately only 30% articles on soil N mineralization were eligible); 2. the study was conducted with the upper soil layer (mostly to the 15 cm soil depth); 3. the condition of soil incubation, in particular the temperature of incubation, was available. For each study, we also extracted the number of replications of soil incubation. Site-specific data were also collected from original articles, including geographic information (latitude, longitude), climatic factors (mean annual temperature, mean annual precipitation), soil physical and chemical properties (the content of clay and sand, soil pH), the substrate features (soil organic C, total soil N, and the ratio of soil C to N), and the characteristics of soil microbial biomass (soil microbial biomass C and the C:N of soil microbial biomass). All data were directly extracted from the text, tables, and figures with the latter extracted by GetData Graph Digitizer (version 2.22). When the climatic factors were unavailable in the original article, they were derived from climatic database (http://worldclim.org/) with grid precision of 0.5 × 0.5° according to geographic information (i.e. latitude and longitude).

The final dataset of Q_min comprised 871 observations from 79 published articles. The dataset encompassed four types of terrestrial ecosystems and included croplands (292 observations), forests (238 observations), grasslands (147 observations), and wetlands (52 observations) (figure 1(a)). The spatial pattern of observations was: 306 from Asia, 160 from Europe, 101 from Australia, 99 from North America, 10 from South America, and the others without geographic coordinates. The range of mean annual temperature varied from −2.2 °C to 30.1 °C, and the range of mean annual precipitation was from 85 to 3100 mm. The soil pH ranged from 3.5 to 9.1, and the content of soil clay from 2.0% to 75.2%.

Zoom In Zoom Out Reset image size

To calculate the soil Q_min, we adjusted the N mineralization to reference temperature (Xu et al 2017), i.e. 25 °C, as the majority (525 out of 871 observations) of soil N mineralization were measured at 25 °C. All N mineralization rates were standardized on the basis of the Q₁₀ of N mineralization, similar to Liu et al (2017), using the following formula:

$$\begin{eqnarray}&&{N}_{{\rm{\min }}2}/{N}_{{\rm{\min }}1}={{Q}_{10}}^{(25-{T}_{1})/10},\end{eqnarray} \tag{ 1 }$$

where N_min1 and N_min2 are the original soil N mineralization and adjusted N mineralization at 25 °C, respectively. T₁ is the measured temperature for N_min1.

We calculated soil Q_min similarly to the metabolic quotient of soil respiration by Xu et al (2017).

$$\begin{eqnarray}&&{Q}_{\min }={N}_{\min 2}/MBN,\end{eqnarray} \tag{ 2 }$$

where Q_min is the metabolic quotient of net soil N mineralization, N_min2 is the net N mineralization rate at 25 °C, and MBN is the soil MBN.

We examined the patterns of soil Q_min and its controlling factors at a global scale using a linear mixed-effect model. Firstly, we tested the patterns of soil Q_min along latitude and longitude, and then we explored the relationships between soil Q_min and the controlling factors. The model was used as:

$$\begin{eqnarray}&&\mathrm{ln}\left({Q}_{\min }\right)={\beta }_{0}+{\beta }_{1}\times \,\mathrm{ln}\,X+{\pi }_{study}+\varepsilon ,\end{eqnarray} \tag{ 3 }$$

where β₀, β₁, π_study and ε are the intercept, slope coefficient, the random effect, and sampling error, respectively. The random effect took the autocorrelation among observations within each 'study' into account. The bivariate relationships between ln (Q_min) and variable were presented as the intercept (β₀) plus coefficient (β₁) times variable plus residuals (ε).

Secondly, the relationship between soil Q_min and each environmental factor was examined in each ecosystem type. All data were standardized using z score normalization in each ecosystem, and the relationships were tested by equation (3). The slope ± its 95% confidence intervals of each relationship were graphically presented. All statistical analyses were conducted with R (version 3.5.0., R Development Core Team, Vienna, Austria).

Thirdly, structural equation modelling (SEM) was used to explore the multivariate relationships between soil Q_min and environmental factors using the normalized data. Initially, we constructed the conceptual SEM, where the climatic factors (mean annual temperature, mean annual precipitation), soil properties (soil pH and the content of soil clay), the substrate (the content of soil N and the ratio of soil C to N), and the properties of soil microbial biomass (soil microbial biomass C and the C:N ratio of soil microbial biomass) might be directly related to soil Q_min. There were also relationships among the environmental factors. For instance, the properties of soil microbial biomass might be influenced by climatic factors and soil properties. The conceptual SEM was tested with the piecewiseSEM package (Lefcheck 2016). To prevent overfitting, we removed variables, e.g. mean annual precipitation, that were not significantly related to soil microbial biomass C, the C:N ratio of soil microbial biomass, and soil Q_min. The optimal SEM was chosen with the lowest Akaike Information Criterion value.

Soil Q_min was the highest at the equator and tended to decrease with the increasing latitude in both Northern (r² = 0.33, P = 0.002; figure 1(b)) and Southern Hemisphere (r² = 0.29, P = 0.756). There was no significant relationship between Q_min and longitude (r² = 0.32, P = 0.319; figure S1 is available online at stacks.iop.org/ERL/15/034005/mmedia).

Soil Q_min increased with mean annual temperature with the slope being 1.29 (r² = 0.32, P = 0.003; figure 2(a)), but did not significantly change with mean annual precipitation (r² = 0.30, P = 0.397; figure 2(b)). Soil Q_min decreased with soil pH (r² = 0.35, P = 0.033; figure 3). Soil Q_min was slightly higher in soils with higher content of sand, and slightly lower with higher contents of clay, but the relationships of Q_min with soil sand or clay contents were not statistically significant (r² = 0.37, 0.39; P = 0.268, 0.559; respectively).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Soil Q_min was significantly related to the properties of the substrate and microbial biomass. Specifically, soil Q_min decreased with the content of soil organic C with the slope being −0.17 (r² = 0.34, P < 0.001; figure 3(d)) and the content of total soil N (r² = 0.37, P = 0.001; figure 3(e)). Moreover, there was a negative relationship between soil Q_min and the ratio of soil C to N (r² = 0.39, P < 0.001; figure 3(f)). Soil Q_min decreased with soil microbial biomass C (r² = 0.30, P < 0.001; figure 4). But soil Q_min increased with the C:N ratio of soil microbial biomass (r² = 0.37, P < 0.001).

Zoom In Zoom Out Reset image size

SEM indicated that mean annual temperature, soil pH, soil microbial biomass C, the C:N ratio of soil microbial biomass had direct effects on soil Q_min at a global scale (figure 5). Specifically, soil Q_min was positively related to mean annual temperature (coefficient = 0.26, P < 0.001) and the C:N ratio of soil microbial biomass (coefficient = 0.39, P < 0.001), while Q_min was negatively related to microbial biomass C (coefficient = −0.20, P = 0.024) and soil pH (coefficient = −0.11, P < 0.001).

Zoom In Zoom Out Reset image size

Mean annual temperature, soil pH, and the ratio of soil C to N also influenced soil Q_min via changing soil microbial biomass C. The effects from mean annual temperature (coefficient = 0.27, P = 0.006) and the ratio of soil C to N (coefficient = 0.28, P < 0.001) on soil microbial C were positive, whereas the effect from soil pH on soil microbial biomass C was negative (coefficient = −0.12, P = 0.046). Moreover, soil microbial biomass C (coefficient = 0.17, P < 0.001) and the content of soil clay (coefficient = 0.08, P = 0.041) positively influenced soil Q_min via changing the C:N ratio of soil microbial biomass.

Among these relationships, the correlation between the C:N ratio of soil microbial biomass and soil Q_min was prominent with standardized coefficient of 0.39. The total effect (the sum of direct and indirect effects) of mean annual temperature on soil Q_min was of the second importance (standardized coefficient = 0.21). The lesser important factors included soil microbial biomass C (total effect = −0.13) and soil pH (total effect = −0.09).

Soil Q_min correlated positively with the C:N ratio of soil microbial biomass in all ecosystem types, with the slope being 0.38 (SE = ±0.075; P < 0.001) in croplands, 0.22 (SE = ±0.061; P < 0.001) in forests, 0.39 (SE = ±0.072; P < 0.001) in grasslands, and 0.50 (SE = ± 0.098; P < 0.001) in wetlands (figure 6). Besides the C:N ratio of soil microbial biomass, the determinants on soil Q_min differed with ecosystem type. In croplands, soil Q_min correlated positively with mean annual temperature (slope = 0.34, SE = ±0.191; P = 0.088) and negatively with soil pH (slope = −0.19, SE = ±0.115; P = 0.087). In forests, soil Q_min correlated significantly with soil microbial biomass C (slope = −0.23, SE = ±0.038; P < 0.001), soil organic C (slope = −0.18, SE = ±0.045; P < 0.001), total soil N (slope = −0.14, SE = ±0.096; P < 0.001), and the ratio of soil C to N (slope = −0.22, SE = ±0.096; P < 0.001). In grasslands, soil Q_min was negatively related to soil microbial biomass C (slope = −0.50, SE = ±0.094; P < 0.001), total soil N (slope = −0.25, SE = ±0.113; P = 0.030), and mean annual precipitation (slope = −0.37, SE = ±0.155; P = 0.021). In wetlands, soil Q_min decreased significantly with the mean annual temperature (slope = 0.35, SE = ±0.141; P = 0.019).

Zoom In Zoom Out Reset image size

The C:N ratio of soil microbial biomass is the paramount factor determining the global variation of soil Q_min, which showed positive relationship with each other. Previous studies found a negative relationship between soil metabolic quotient of soil respiration and microbial C:P, which indicates the significant influences of microbial stoichiometry on soil metabolic quotient of soil respiration (Hartman and Richardson 2013). As pointed out by Cleveland and Liptzin (2007), soil microbial stoichiometry could mirror the relative microbial nutrient limitation. At a global scale, soil microbial biomass is relatively restricted by N in the terrestrial ecosystem (Cleveland and Liptzin 2007). Therefore, microbes increase N metabolism to gain N. Greater exudation of enzymes used to obtain N from the environment may be a strategy for microbes when available N is limited. Urease is negatively correlated with the contents of soil inorganic N (Ajwa et al 1999). Moreover, the arginine deaminase, alanyl aminopeptidase, lysyl–alanyl aminopeptidase increase under low N treatment (Enowashu et al 2009). The exudation of enzymes might consume the N of microbial biomass, which, in turn, result in higher C:N ratio of soil microbial biomass. When N is limiting, the metabolism of soil microbes should be more sensitive to N availability. Consistent with our assumption, the catabolic activity of soil microbial biomass increases with the additional N input in the natural ecosystem (i.e. grassland in Cedar Creek) (Fierer et al 2012). In addition, soil physical property (i.e. the content of clay) and microbial biomass influenced soil Q_min via changing the C:N ratio of soil microbial biomass (figure 5), indicating that the properties of soil microbial biomass are the critical bolt for soil Q_min.

Mean annual temperature is the second important controlling factor on soil Q_min on a global scale. The previous experiment also shows that the microbial metabolic quotient of soil respiration substantially increases with temperature (Xu et al 2006, Nazaries et al 2015). This phenomenon is ascribed to the increasing costs of sustaining microbial activities under higher temperature (Alvarez et al 1995), which is verified by decreasing C utilization efficiency under higher temperature (65.4% for 15 °C versus 59.5% for 25 °C) (Steinweg et al 2008). Additionally, the activities of soil extracellular enzymes accelerate under higher temperature. For example, the activity of soil urease significantly increases by 33% in a natural forest when the soil temperature rises by 0.55 °C (Xu et al 2010). The results from 16 experiments across the United States revealed that experimental warming significantly increases proteolytic activity by 18% in organic horizons (Brzostek et al 2012). When the soil temperature is increased by 0.98 °C, soil N-acetylglucosaminidase increases by 5% or even up to 21% (Zhou et al 2013). These suggest that increasing mean annual temperature under global climate change might accelerate soil N cycling via motivating soil Q_min.

The size of soil microbial biomass C negatively influenced soil Q_min, which might attribute to three reasons. First, given the same amount of soil substrate, with more soil microbes, the efficiency of decomposing substrate might become less. Second, soil microorganisms are highly diverse, and they could be grouped into active, potentially active, and dormant microbial states (Blagodatskaya and Kuzyakov 2013). In general, majority of soil microbial biomass cannot participate in ecosystem processes (Blagodatskaya and Kuzyakov 2013). This might account for the negative relationship between soil microbial biomass C and Q_min. Third, the bigger soil microbial community might arouse a greater competition with one another. Given the scarcity of resources in most soil ecosystems, more soil microbes might lead to fierce competition (Ghoul and Mitri 2016). The microbes might excrete the secondary metabolites, i.e. antibiotics, to compete with one another. This could consume the energy of microorganism. It was reported that 5%–10% of genomes link to secondary metabolites in actinomycetes (Nett et al 2009). Consistent with the negative relationship between soil Q_min and microbial biomass C, soil microbial metabolism of respiration with ecosystem succession shows the contrary trends in comparison with that of the microbial biomass (Wardle and Ghani 2018).

Soil pH is the second negative controller on global soil Q_min. Soil pH could negatively influence the activities of enzymes taking part in N cycling. A global synthesis revealed that β-N-acetylglucosaminidase decreases (slope = −0.54) with pH changing from 4 to 9 (Sinsabaugh et al 2008). The activity of urease decreases when soil pH increased from 4.5 to 8.5 (Singh and Nye 1984). In addition, previous studies found that some protease is optimal under acidic conditions (Kamimura and Hayano 2000).

Although the positive relationships between soil Q_min and C:N ratio of soil microbial biomass were consistently found across all ecosystem types, there were different drivers, other than the C:N ratio of soil microbial biomass, on soil Q_min in different ecosystem types. In forests and grasslands, soil Q_min was correlated negatively with soil microbial biomass C. The soils in forests (644 μg g⁻¹ soil) and grasslands (416 μg g⁻¹ soil) usually possess more microbial biomass C than adjacent croplands (313 μg g⁻¹ soil) (Prasad et al 1995). It was reported that soil microbial biomass increases along the forest succession (Wardle and Ghani 2018), the competition among microorganisms likely increases as well, indicating an increasing negative regulation of microbial biomass on Q_min. In grasslands, soil Q_min also significantly correlated with mean annual precipitation. The water shortage in grasslands might limit soil N cycling (Reichmann et al 2013, Hu et al 2015). In wetlands, soil Q_min significantly accelerated with higher mean annual temperature. The higher temperature can expedite the microbial metabolism (Tveit et al 2015) and the activities of soil extracellular enzymes (e.g. urease increased by 9%; (Bai et al 2018)). Soil N mineralization generally occurs under aerobic conditions, the anaerobic condition of wetland might partially mask the other relationship between soil Q_min and soil microbial properties (i.e. microbial biomass C).

In croplands, the insignificant relationship between soil Q_min and microbial biomass C may be due to that anthropogenic disturbance significantly impacts soil microbial biomass (Helgason et al 2010, Da Silva et al 2014). The croplands exposed to greatest anthropogenic disturbances, such as tillage, fertilization, etc, may directly affect the size of soil microbial biomass. For instance, soil microbial biomass C decrease by 19% under tillage (Zuber and Villamil 2016), wherein the relationship between soil Q_min and microbial biomass should be enhanced. However, the disturbance could break the competition among the community (Huston 1979). For example, tillage could weaken the competition within the soil microbial community (Bailey and Lazarovits 2003). Particularly, fertilization (e.g. urea) provide more substrate to mineralize by microbial biomass, so the negative relationship between soil Q_min and microbial biomass weakened or even disappeared (figure 6). However, soil pH exhibited marginally significantly negative influence on soil Q_min. The fertilization could elicit soil acidification (Guo et al 2010). The acidification could positively affect soil Q_min by changing the activities of soil enzymes in N cycling. For instance, soil urease activity decreases with higher soil pH (Singh and Nye 1984).

The changes in soil microbial stoichiometry under global change will play an important role in soil N cycling via changing soil Q_min. As reported, the ratio of C to N in bacteria averages 6.5, but that in fungi is 5–17 (Cleveland and Liptzin 2007), therefore, any shift in the ratio of fungi to bacteria under global change could eventually change soil Q_min. The results from a 3 year warming experiment showed that the ratio of fungi to bacteria increases 22%–63% (Zhang et al 2005). The C:N ratio of soil microbial biomass is also increased by 0.8%–6.2% in an alpine meadow under warming (Fu et al 2012). A 70% increase of N mineralization in arctic zones under warming (Aerts et al 2006) is partially ascribed to the changes in microbial biomass stoichiometry and increases of enzymatic activities. Additionally, the changes in the stoichiometry of soil microbial biomass, which is significantly decreased under N deposition (Bell et al 2010), will negatively affect soil Q_min and subsequently reduce the N mineralization. Moreover, the ratio of fungi to bacteria increases by 8% at elevated CO₂ (Guenet et al 2012), which likely results in higher soil Q_min under increasing atmospheric CO₂ concentrations.

More severe soil acidification under N deposition and/or N addition might elicit significant changes in soil N cycling. Apart from the direct effect of soil pH on soil Q_min, the indirect effect via changing the size of soil microbial biomass is also important. Soil pH generally dominates the distribution patterns of bacteria (Fierer and Jackson 2006) and fungi at the global scale (Tedersoo et al 2014). The bacterial community is more sensitive to soil pH than fungi (Rousk et al 2010). Moreover, the fungal: bacterial growth is significantly decreased with soil pH ranging from 3 to 8 (Rousk et al 2011). The meta-analysis showed that soil pH has decreased on average by 0.26 in natural ecosystems (Tian and Niu 2015) and 0.5 in croplands (Guo et al 2010). To meet the food demand in the context of population growth, the fertilizer inputs might not decrease in the next century (Erisman et al 2008), which will strengthen soil acidification. Eventually, soil acidification would fortify the microbial ability of soil N mineralization.

Current soil N models are built on either the dynamics of N pool (Wang et al 2004) or the empirical response of soil mineralization to temperature (Paul et al 2002). However, the microbial mechanisms have not been well presented in models to predict soil N availability. Soil microbial properties, like the size of microbial biomass and the stoichiometry, are very sensitive to environmental changes (Buchkowski et al 2015). With mounting data on the stoichiometry of microbial biomass (Xu et al 2013), it is urgent to incorporate soil microbial mechanisms into global N cycling model (Wieder et al 2015a).

We are aware that changes in soil microbial community and extracellular enzymes are also important for soil N cycling. However, due to data paucity, they are unable to be included in this study. The projection of soil biogeochemical processes will be more precise when the responses of soil microbial communities are taken into consideration (Karhu et al 2014). Soil extracellular enzymes are the direct participants in biogeochemical processes, which have drawn growing attentions in biogeochemical processes (Chen et al 2018), and there are many enzymes in N cycling, such as urease, proteinase, amidases, and deaminases, etc However, there is a shortage of data to reveal the influences of soil enzymes on soil Q_min on a global scale. In addition, soil Q_min was calculated using the data from laboratory incubation, which might be higher than that calculated using in situ data, but they could show a similar relationship at the global scale (r² = 0.87) (Xu et al 2017). The relationships between soil Q_min and the impacting factors remain to be tested when there would be enough data from field studies.