Bioclimate and arbuscular mycorrhizal fungi regulate continental biogeographic variations in effect of nitrogen deposition on the temperature sensitivity of soil organic carbon decomposition

2.1 Study area

We collected 92 soil samples from seven sites of forest and grassland ecosystems with various climate and latitude (Table S1). These sites have distinct soil properties and climate conditions with a MAT range of −2.4 to 21.7°C and a mean annual precipitation (MAP) range of 350 mm to 1,700 mm. Long-term simulated N deposition experiments have been established more than 5 years. N was applied as ammonium nitrate (NH₄NO₃) and was dissolved in water and sprayed at some interval with a backpack sprayer. The control plot only received the same volume of water. At all sites except for HB, the experiments included three treatments: control (no fertilization, CT), low N addition (LN), and high N addition (HN). Detailed information about the long-term simulated N deposition experiment at each site is provided in Table S1. Bioclimatic variables were calculated from the monthly temperature and precipitation values to generate biologically meaningful variables. For each site, 11 standardized climatic variables (Table S2) for the current climate were obtained from the Worldclim database (www.worldclim.org) (Delgado-Baquerizo et al., 2017).

2.2 Soil sampling and analysis

Soil samples were collected from each site in August, 2017. At each site, we selected 9–15 plots with 100–150 m², and each plot was 50–100 m far away. In each plot, after removing the surface litter layer, the top mineral soil with 10 cm depth was randomly collected from 8–10 locations using a metal corer with a 5 cm diameter and then gently mixed to a composite sample. In this study, 92 composite soil samples were amalgamated. All soil samples were transported in a low-temperature incubator with ice (<5°C) to the laboratory within 24 hr after collection and then passed through a 2 mm mesh to eliminate residual organic matter (i.e., decomposed leaf litter and roots). Some of each sample were used for incubation to determine soil microbial respiration. The rest of each sample were used for the determination of soil chemical and microbial properties.

Soil organic C and total N concentrations of soil samples were measured using a C/N analyzer (Elementar, Germany). Labile organic C (LOC), which is oxidized by 333 mM potassium permanganate, was measured in accordance with the method described by Blair, Lefroy, and Lisle (1995). Soil total P was measured colourimetrically, whereas soil mineral N (sum of ammonium and nitrate N) was extracted using 2 M KCl solution and determined by colourimetry (Lu, 2000). Available soil P was analyzed colourimetrically through the molybdate blue method after the soil was extracted with 1 M NH₄F solution. Soil pH was determined using a pH meter in a 1:2.5 (weight: volume) mixture of soil and deionized H₂O. The soil texture (sand, silt, and clay contents) was analyzed using the pipette method (Lu, 2000). The concentrations of exchangeable Na⁺, Ca²⁺, K⁺, and Mg²⁺ and other variables were all analyzed in accordance with the standard protocols reported by Wang, Liu, and Tian (2018).

Soil microbial community was measured using Illumina sequencing after extracting genomic NDA, and detailed information can be found in the study by Liu, Wang, Deng, Tian, and Wang (2018). Phospholipid fatty acids (PLFAs) were determined in accordance with the method described by White and Ringelberg (1998). Soil amino sugars were extracted as described by Zhang and Amelung (1996). As an index of fungal and bacterial residues, bacterial- and fungal-derived C were calculated as follows. Bacterial-derived C (mg g⁻¹ dry weight) was calculated as an index for bacterial residues by multiplying the content of muramic acid in mg g⁻¹ dry weight by 45 (Appuhn & Joergensen, 2006). Fungal-derived C (mg g⁻¹ dry weight) was calculated by subtracting bacterial glucosamine from total glucosamine as an index for fungal residues, assuming that MurN and GluN occur at a 1-to-2 M ratio in bacterial cells: mg glucosamine g⁻¹ dry weight = (mmol glucosamine −2 × mmol muramic acid) × 179.2 × 9, where 179.2 is the molecular weight of glucosamine and 9 is the conversion value of fungal glucosamine to fungal C (Appuhn & Joergensen, 2006).

2.3 Soil incubation and Q₁₀ calculation

The soil samples from all sites were adjusted to 60% of water-holding capacity (WHC) and then preincubated for 7 days to minimize the 'pulse effect'. Each soil sample was split into two-part subsamples of 20 g (dried weight) and placed into 80 mL glass culture flasks. All soil subsamples were placed inside incubators and incubated over a 30-day incubation period. One part was at MAT and another part was at MAT plus 5°C. The basic incubation temperature was 10°C when the MAT of soil original sites was close to or below 10°C. In this experiment, we used a novel incubation method with periodically varying temperatures based on the MAT of soil original sites. The incubator (LRH-250, Yiheng Scientific Instrument Co., Ltd., Shanghai, China) automatically regulated its inner temperature that gradually changed between the designed incubation temperature plus 5°C and minus 5°C within 24 hr. As a result, the time-weighted average temperature was equal to the designed incubation temperature. The temperature of inner incubator increased with the rate of 1°C every 1.2 hr to the designed incubation temperature, and then decreased with the rate of 1°C every 1.2 hr. Further information about this incubation methodology is provided in Figure S1. During incubation, all incubation flasks were open to ensure enough O₂ for soil microorganisms, and soil samples were maintained with 60% WHC by adding distilled water. The rate of CO₂ emission was measured using an infrared gas analyzer (Li-Cor 820). Before measuring CO₂ emission, the incubator temperature was adjusted to the designed temperature and maintained for 12 hr (Chen et al., 2010). Before measurement, we connected the gas analyzer to each 80 mL incubation flask in a close-loop configuration. Each measurement generally can be finished within 5 min. The Q₁₀ of SOC decomposition was calculated by the following equation: Q₁₀ = (R_M5/R_M)^10/5, where R_M5 and R_M are the rates of CO₂ emission at MAT plus 5°C and MAT, respectively, and 5 was the temperature difference between MAT plus 5°C and MAT.

2.4 Compiling data

To obtain a generally biogeographic pattern of N deposition effect on Q₁₀, we also collected and compiled the published data from reviewed literature published in 1990–2018. We collected journal articles by searching the Web of Science and China National Knowledge Infrastructure databases. The search parameters were limited to papers which titles, abstracts and keywords referred to 'heterotrophic respiration (Rh)', 'soil organic carbon', or 'matter decomposition' 'CO₂ emission' or 'flux', 'temperature', and 'N addition' or 'fertilization' or 'deposition'. Of the articles retrieved by the search, those conducted in forest and grassland ecosystems were selected. Furthermore, these articles were identified based the following criteria: (a) Rh and corresponding soil temperature were repeatedly measured for at least one season when the experiment was conducted in field, or soils used in the laboratory incubation experiment were collected from long-term N addition plots; (b) the application rate and form of N were clearly given; and (c) Q₁₀ was calculated as the equations (Q₁₀ = e^10β, Rs = ae^βT) for the experiment conducted in field. In total, 27 studies (Table S3) met these criteria and were used in this study. DATA THIEF software (GetData Graph Digitizer) was used to extract data from figures.

2.5 Data analysis

N deposition effect on Q₁₀ and soil variables was calculated as the ratio of the value in N deposition plots to that in control plots, which was used to assess the effects of nutrient addition on soil microbes (Leff et al., 2015). At each site, significances in Q₁₀ values between the control and N deposition treatments and N deposition effect on Q₁₀ in low and high N deposition treatments were analyzed using t-test. A non-parametric test (Kruskal-Wallis) was used, followed by Dunn's post hoc tests to compare means between sites and between N deposition types. Pearson's correlation analysis was conducted between N deposition effects on Q₁₀ and latitude. Nitrogen effect is the average values of low and high N levels for our measured data. Furthermore, multiple relationships among N deposition effects on Q₁₀, soil properties and microbial communities, and bioclimatic variables were performed to further reveal the key regulating factors of Q₁₀ response to increased N input. Microbial community composition was related to each environmental factor by partial Mantel tests. The significance of each variable was indicated by the relation coefficient with N deposition effects on Q₁₀.

Variation partitioning modeling was performed to assess the relative importance of three groups, namely, bioclimate, soil properties, and microbial communities, in driving continental variations in N deposition effects on Q₁₀. We used the 'forward.sel' function to avoid redundancy and multicollinearity in variation partitioning analysis. We further used a random forest model to quantify the relative contribution of some important factors in regulating the biogeographic variation in N deposition effects on Q₁₀. We first conducted the 'select.forward' procedure using the Vegan package to select the best variables of bioclimatic variables, soil properties, and microbial communities, and then permutations with 999-times to test the significance (p < .05) of variables in predicting the N deposition effects on Q_10. All statistical analyses were performed in R 3.5.1.

3.1 Biogeographic variation in effects of N deposition on Q₁₀

Across all soils, the Q₁₀ values of SOC decomposition had large variations, ranging from 0.69 to 6.01 (Figure 1). The mean Q₁₀ values were 2.39 and 2.64 in the control and N addition treatments, respectively. Nitrogen deposition effects on Q₁₀ were greater than 1.0 across all sites (Figures 1 and 2), although they had large site variations. Nitrogen deposition significantly increased Q₁₀ at the HS, JG, HB, and DL sites but not at other sites, indicating that responses of Q₁₀ to N deposition had strong site-dependence. Furthermore, the magnitude of N deposition effect on Q₁₀ was significantly different between LN and HN at the JG and DL sites (Figure 2). Together with the data collected from the previous experiments, N deposition effect on Q₁₀ was linearly and negatively related to latitude (Figure 3), suggesting that elevated N input would exert great positive effect on the response of SOC decomposition to elevated temperature in warm terrestrial ecosystems at low latitudes.

3.2 Controlling factors of N deposition effects on Q₁₀

Multiple-relationship analysis results revealed strong connections among bioclimatic factors and N deposition effects on biotic and edaphic factors (Figure 4), suggesting that their interactions should be considered when predicting the response of SOC decomposition Q₁₀ to elevated N input. Among bioclimatic variables, BIO3 (r = −0.31, p < .05), BIO4 (r = −0.35, p < .01), and BIO8 (r = −0.33, p < .05) negatively correlated with N deposition effect on the Q₁₀ of SOC decomposition (Figure 4). For edaphic variables, the magnitudes of N deposition effect on exchangeable soil Ca (i.e., Ca²⁺) (r = −0.40, p < .01) and pH (r = −0.31, p < .05) were negatively but those for available K (r = 0.43, p < .01) and clay contents (r = 0.295, p < .05) were positively related to N deposition effect on Q₁₀ (Figure 4 and Figure S2a). Responses of fungal biomass measured by PLFAs (r = −0.28, p < .05) and soil C:N ratio (r = −0.28, p < .05) to N deposition were negatively related to N deposition effect on Q₁₀ (Figure S2b), indicating that the response of fungal biomass and soil C:N ratio to N deposition are also important in regulating the effects of N deposition on Q₁₀.

Variation partitioning analysis results demonstrated that all variables measured in this study totally explained 50.1% of the spatial variation in N deposition effect on Q₁₀ (Figure 5a). Among all factors, bioclimate (16.0%) had greater unique explanation for spatial variation in N deposition effect on Q₁₀ than soil (11.8%) and microbial properties (6.3%). The results of random forest model analysis further showed that among bioclimatic factors, BIO4, BIO8, and BIO3 were more important and contributed 16.5, 10.8, and 9.5% relative importance of the spatial variation regulating N deposition effect on Q₁₀, respectively (Figure 5b). For soil properties, the magnitude of effects of N deposition on soil available K and exchangeable Ca had higher predictive power for N deposition effects on Q₁₀, contributing 18.5% and 7.2% relative importance, respectively. Glomeromycota was the best microbial predictors and contributed 10.9% relative importance of the spatial variation regulating N deposition effect on Q₁₀, respectively. In addition, Acidobacteria was also important in regulating the variation in N deposition effect on Q₁₀.

4.1 Biogeographic variation in N deposition effects on temperature sensitivity

In general, the magnitude of N deposition effects on Q₁₀ was greater than 1.0 across all sites (Figure 1), which was contrary to our hypothesis and some previous studies (Mo et al., 2008; Zang et al., 2020). Together with the observation of Coucheney et al. (2013), our results suggest that N deposition increased the sensitivity of SOC decomposition to elevated temperature. The decreased proportion of LOC to SOC pool in our study (decreasing on average by 20.3%, data not shown) and other studies (e.g., Guo et al., 2017; Janssens et al., 2010) supported the conclusion that the enhanced Q₁₀ under long-term N input can be well explained by the change in SOC quality characterized with C complexities. This is consistent with the C quality-temperature hypothesis that SOC with higher recalcitrance has higher temperature sensitivity (Davidson & Janssens, 2006; Jia et al., 2020). However, from the first study to quantify the biogeographic variation in the effects of N deposition on Q₁₀ of SOC decomposition at the continental scale, we found that N deposition effects on Q₁₀ had large variations with a range of 0.46–3.60 (Figures 1 and 2) and demonstrated strong site dependence. The strong site dependence was not surprising, given the broad range in climate, soil characteristics, and plant community composition across sites that affect Q₁₀ (e.g., Conant et al., 2011; Davidson & Janssens, 2006; Wang, Liu, Wang, et al., 2018). The N deposition effect on Q₁₀ was significantly negatively correlated with the magnitude of the response of edaphic variables and microbial properties to increased N input (Figure 4), helping to explain site-to-site variability in shifts in Q₁₀ of SOC decomposition and suggesting that sites where N deposition had the slightest impacts on edaphic variables and microbial properties were also the sites that had the strongest N effects on Q₁₀.

Overall, N deposition effect on Q₁₀ was negatively correlated with latitude (Figure 3), which was contrary to the spatial pattern of Q₁₀ with latitude (e.g., Johnston & Sibly, 2018; Li et al., 2020; Peng, Piao, Wang, Sun, & Shen, 2009), indicating that N deposition alters the response of Q₁₀ to elevated temperature. The negative correlation also suggests that N deposition will have greater positive influence on the response of SOC decomposition to elevated temperature in the low-latitude ecosystems than the high-latitude ecosystems, that is to say, soils in the low-latitude ecosystems will have higher risk of SOC losses than those in the high-latitude ecosystems in response to the combination of climate warming and atmospheric N deposition (Wagai et al., 2013). Nitrogen limitation for plants and soil microorganisms in high-latitude ecosystems is lessened or even disappeared after N input and litter C:N ratios decrease, whereas P limitation is exacerbated by N input in the low-latitude ecosystems where soil P availability is generally a limiting factor but N may have already been saturated (Khan, Mulvaney, Ellsworth, & Boast, 2007). Therefore, shifts in nutrient limitation for soil microorganisms caused the linear pattern of responses of Q₁₀ to N deposition with latitude.

4.2 Factors regulating N deposition effect on temperature sensitivity

As our second hypothesis, results of variation partitioning and random forest model analysis demonstrated that the bioclimate, particularly BIO3, BIO4, and BiO8 related to temperature, had more importance in regulating N deposition effect on Q₁₀ (Figure 5), explaining 33.3% of the total variations. This study was the first to find the role of bioclimate variables in regulating N deposition effect on Q₁₀ across sites. Bioclimate as an important abiotic factor similar to temperature directly affects the Q₁₀ of SOC decomposition (e.g., Davidson & Janssens, 2006). Bioclimate can also indirectly influence Q₁₀ and its response to N deposition through its influence on the amount and quality of SOC, soil properties, and microbial community by modifying or even altering plant community structure (e.g., Clark & Tilman, 2008; Janssens et al., 2010; Leff et al., 2015; Ramirez, Craine, & Fierer, 2012). This is because plants with diverse traits can provide organic matter input to soil as litters with differing C and energy resources (e.g., Fontaine et al., 2007; Pascault et al., 2013) and nutrient stoichiometries (Delgado-Baquerizo et al., 2017). These differences can lead to variations in soil properties, such as pH and SOC quality, and microbial properties, such as community composition and activity (Janssens et al., 2010; Ramirez et al., 2012; Treseder, 2008), and consequently exert strong effects on response of Q₁₀ of SOC decomposition to N deposition.

Edaphic variables explained 23.7% of the total variation in response of Q₁₀ to N deposition (Figure 5a), suggesting that soil properties had important contribution to the spatial variation in N deposition effect on Q₁₀. Furthermore, the magnitude of response of exchangeable soil Ca²⁺ to N deposition was found to be a better edaphic predictor (Figure 5b). Increased N input to terrestrial ecosystems may promote the shortage of Ca via leaching and depletion (Hynicka, Pett-Ridge, & Perakis, 2016; Tafazoli, Hojjat, Jalilvand, & Lamersdorf, 2019), which can impact plant productivity and SOC cycling (Battles, Fahey, Driscoll, Blum, & Johnson, 2014; Reich et al., 2005), although the effects of N deposition on Ca²⁺ leaching and depletion are related to initial N levels in ecosystems (Hynicka et al., 2016). Furthermore, exchangeable soil Ca²⁺ can indirectly exert their influence on the size and activity of the microbial community through modifying soil physicochemical environment and thereby control SOC decomposition (e.g., Rakhsh, Golchin, Agha, & Alamdari, 2017). For example, exchangeable soil Ca²⁺ facilitates SOC stabilization by clay flocculation and electrostatic bridging between clay surfaces and SOC (e.g., Kögel-Knabber & Kleber, 2011). Therefore, decrease in exchangeable soil Ca²⁺ content would increase the microbial availability of SOC.

As expected, we found that microbial properties totally and uniquely explained 14.3% and 6.3% of the spatial variations in N deposition effects on Q₁₀, respectively [Figure 5(a)]. This study was the first to clarify the importance of soil microbial properties in regulating the response of Q₁₀ to N deposition across sites, although some studies have demonstrated that soil microbial community can partly explain the spatial variation in Q₁₀ of soil respiration (Johnston & Sibly, 2018; Walker et al., 2018; Wang et al., 2018b). The result was not surprising because soil microorganisms govern SOC cycling and N deposition can change their activity and composition (Hasselquist et al., 2012; Johnson, Wilson, Bowker, Wilson, & Miller, 2010; Leff et al., 2015; Wei et al., 2019). We also found that Glomeromycota, a group of arbuscular mycorrhizal fungi (Redecker & Raab, 2006), is the best microbial predictor and contributes 10.9% relative importance of the spatial variations regulating N deposition effect on Q₁₀ (Figure 5(b)). Increased N input typically decreases the relative abundance and biomass of Glomeromycota (e.g., Janssens et al., 2010; Leff et al., 2015; Ramirez et al., 2012; Treseder, 2008). This decrease in the biomass and abundance of arbuscular mycorrhizal fungi reduces the stability of soil aggregation and SOC because arbuscular mycorrhizal fungi have positive effects on soil aggregate formation through their extensive hyphae system (e.g., Lehmann & Kleber, 2015). Therefore, decrease in arbuscular mycorrhizal fungi constituted a mechanism regulating the Q₁₀ of SOC decomposition in response to N deposition, which is indirectly supported by significant correlations between SOC decomposition and arbuscular mycorrhizal fungi following nutrient addition (Guo et al., 2017).

In summary, our findings from cross-site experiments using the novel incubation method together with literature data provide a strong evidence that the effects of N deposition on the temperature response of SOC decomposition varied with climate-associate biogeographic variation. Our study represented the first attempt to empirically assess whether N deposition effects on Q₁₀ show generalizable patterns across a wide range of climatic and edaphic environments and confirmed their existence, despite large cross-site differences in N deposition effect on Q₁₀. Furthermore, we revealed the potential mechanisms underlying the spatial variation in N deposition effect on Q₁₀ and quantified the relative importance of bioclimatic, edaphic, and microbial variables in regulating the variation in N deposition effect on Q₁₀. In particular, this study found for the first time that BIO4, exchangeable soil Ca²⁺ and arbuscular mycorrhizal fungi as indicated by Glomeromycota are important predictors of the spatial variation in N deposition effects on Q₁₀. In consideration that the biogeographic variation in response of Q₁₀ to increased N input is an important source of uncertainties in quantifying terrestrial C cycle, this variation should be considered in carbon-climate models to decrease the prediction uncertainties of SOC dynamics and its feedback to global warming in the increasing atmospheric N deposition context.