Nitrogen addition alters elemental stoichiometry within soil aggregates in a temperate steppe

Ongoing increases in anthropogenic nitrogen (N) inputs have largely affected soil carbon (C) and nutrient cycling in most terrestrial ecosystems. Numerous studies have concerned the effects of elevated N inputs on soil dissolved organic carbon (DOC), dissolved inorganic N (DIN), available phosphorus (AP), exchangeable calcium (Ca) and magnesium (Mg), and available iron (Fe) and manganese (Mn). However, few have emphasized the stoichiometric traits of these soil parameters, especially within different soil aggregate fractions. In a semiarid grassland of Inner Mongolia, we studied the effect of N addition on the ratios of DOC : DIN, DOC : AP, DIN : AP, exchangeable Ca : Mg, available Fe : Mn within three soil aggregate classes of large macroaggregates (> 2000 μm), small macroaggregates (250–2000 μm), and microaggregates (< 250 μm). Elevated N inputs significantly decreased the DOC : DIN ratio within three soil aggregates. The soil DOC : AP ratio significantly decreased along with increasing N gradients within large macroaggregates and microaggregates. Nitrogen significantly decreased the ratio of exchangeable Ca : Mg within soil macroaggregates. The ratio of available Fe : Mn decreased with N addition within three soil aggregate classes. Alteration of elemental stoichiometry within soil fractions that are characterized by different nutrient retention capacity will influence the chemical composition of soil microorganisms and plant quality.


Introduction
Stoichiometric relationships of soil nutrients are reliable indicators of nutrient availabilities for plants and soil microorganisms (Cleveland and Liptzin, 2007). Soil carbon:nitrogen:phosphorus (C:N:P) ratios play essential roles in shaping plant and microbial stoichiometry which reveals nutrient allocations, life history strategies, and 5 physiological adjustment to climate change (Elser et al., 2010). For instance, enhanced N inputs resulted in higher soil N availability (Wei et al., 2013) and consequentially decrease of foliar C:N and increase of foliar N:P ratio (Han et al., 2014). Nitrogen addition was suggested to cause divergent effects on plant C:P ratio which declined at species-level but showed no change at the community-level (Han et 10 al., 2014). Plant species that are stoichiometric N:P flexibility may have advantage over these possessing strict stoichiometric homeostasis under environmental changes (Sardans and Peñuelas, 2012). Leaves with high nutrient concentrations (both N and P), which is relatively lower C:N and C:P ratios tend to be short lived with high specific leaf area (Wright et al., 2004) as well as high photosynthetic capacity and 15 dark respiration rates (Reich et al., 2008). Local nutrient availability from soils is suggested to be one of the most important contributors to the wide variation of terrestrial plants in C:N:P ratios (i.e. plant are what they root in) (Elser, 2010). Soil exchangeable calcium (Ca) and magnesium (Mg) and available iron (Fe) and manganese (Mn) are critical metal nutrients for plant growth, microbial activity, and 20 ecosystem health (Lucas et al., 2011). The ratio of exchangeable Ca:Mg indicates the relative availability of these two ions and is reported to influence clay dispersion and 4 surface sealing processes (Dontsova and Norton, 2002). Maintaining proper available Fe:Mn ratio is essential for plant health as lower ratio might suggest plants suffering from Fe deficiency and Mn toxicity (Hodges, 2010). However, the stoichiometric qualification of soil DOC and available nutrients is less known than plant stoichiometry. 5 Ongoing increases in anthropogenic N inputs have largely affected soil C and nutrient cycles in most terrestrial ecosystems Neff et al., 2002).
Under N enrichment, higher soil available N (sum of nitrate and ammonium) has been documented to be positively (Wei et al., 2013), negatively (Heffernan and Fisher, 2012) or even neutrally (Liu and Greaver, 2010) related to dissolved organic C (DOC), 10 which is an important linkage for plant-soil-microbe system in natural ecosystems.
Anthropogenic increases in reactive N would also shift ecosystem nutrient limitation from N to P by stimulating P uptake by plants (Menge and Field, 2007). Moreover, soil acidification caused by N enrichment could help desorption and dissolution of soil inorganic P serving as an essential P source for plants, especially in calcareous 15 soils (Lajtha and Bloomer, 1988;Tunesi et al., 1999). Soil pH is widely recognized as one of the most influential factors that regulates nutrient bioavailability (Kemmitt et al., 2006). For example, Piccolo (2001) suggested that drop of soil pH elevated solubility of soil humic substances. Soil pH is also suggested to influence soil C:N ratio via changes in plant litter quality (Schmidt, 1982) and increase soil available P 20 (AP) through promoting phosphomonoesterase activity (Hogan et al., 2010). Thus, N enrichment might influence soil available C:N:P stoichiometry through the alteration 5 of soil pH, and hence aggravate or alleviate plant and microbial C and nutrient limitations (Elser, 2010). Enhanced N inputs causes leaching of soil exchangeable Ca and Mg and activation of soil available Fe and Mn as a result of soil acidification which will lead to nutrient imbalance in soils (Katou, 2002;Malhi et al., 1998). For example, N deposition decreases the ratio of exchangeable Ca:Mg as preferential 5 weathering of soil exchangeable Ca relative to Mg (Lu et al., 2014). Iron deficiency chlorosis has been found in calcareous soils in grassland of America (Rogovska et al., 2007) and Mn toxicity may occur after soil acidification companying with leaching of base cations (Lynch and Clair, 2004). Thus, available Fe:Mn ratio will be useful in indicating relative availability of Fe and Mn as affected by N addition. Under 10 anthropogenic N enrichment scenarios, clarifying stoichiometric traits of soil available elements would promote our understanding of soil nutrient imbalance as caused by environmental changes.
Physical protection by soil aggregation is a main mechanism of soil organic matter (SOM) stabilization . Oxygen and water diffusion rates is 15 generally parallel with soil aggregate sizes as being determined by pore space (Young and Ritz, 2000;Baker et al., 2007). Consequently, available nutrients, like N and P tend to distribute in aggregates of larger size as more favorable microclimate for microbial activity herein (Dorodnikov et al., 2009;Wu, 2012). Smaller aggregate-size classes are confirmed as a preferential site for the physical stabilization of SOM and 20 nutrient retention (Fonte et al., 2014). Higher clay contents in microaggregates tend to provide more surface area and binding sites to retain more exchangeable Ca and Mg 6 (Oorts et al., 2003). Within microaggregates, more microbial-processed SOM will also chelate or complex larger amount of base cations and micronutrients (Lü et al., 2016). Even though numerous studies have focused on C and nutrients cycling in aggregate scale (Fonte et al., 2014;Six and Paustian, 2014;Wright, 2009), stoichiometric traits of DOC and available nutrients within aggregates are still 5 unclear.
Semiarid steppes of Inner Mongolia in northern China, playing a vital role in serving environmental health, regional economy, and the global C cycle, have undergone degradation during the last 50 years' increased stocking rates and static grazing management practices (Kang et al., 2007). Non-physically protected SOM 10 pools, which can be a majority of SOC and nutrients in these sandy grassland systems, are however, inherently vulnerable to environmental changes (Creamer et al., 2011), like elevated N inputs. A 7-year N field manipulation experiment has already showed significant aboveground changes in species turnover rates, plant community composition, and community stability (Xu et al., 2012). Microaggregates (< 250 μm) 15 had the highest dissolved inorganic N (DIN) concentration as compared to large (> 2000 μm) and small macroaggregates (250-2000 μm) (see Wang et al., 2015a) which was not the case for both DOC (see Wang et al., 2015b) and AP .
Nitrogen addition significantly increased DIN (see Wang et al., 2015a) and available Fe and Mn but decreased exchangeable Ca and Mg across three soil fractions (not for 20 Ca in microaggregates) (Wang et al., 2017). We analyzed stoichiometric traits of DOC, DIN, AP, exchangeable Ca and Mg, and available Fe and Mn within soil aggregates in 7 this semi-arid grassland. We hypothesize that microaggregates had lower DOC:DIN and higher DIN:AP ratios as better N retention capacity in smaller-size aggregates.
We also hypothesized that N addition would decrease DOC:DIN while increase DOC:AP and DIN:AP ratios as higher anthropogenic inputs of inorganic N and intensive removal of available P from soil by plant uptake. As preferential weathering 5 of Ca relative Mg during soil acidification, we expected to detect significant decrease

Study site and experimental design
The study site (42°02′27″N, 116°17′59″E, elevation 1,324 m a.s.l) was located in Duolun County, a semi-arid area in Inner Mongolia, China. Mean annual temperature 15 is 2.1 º C, ranging from 17.8 º C in January to 18.8 º C in July and average precipitation is 379.4 mm yr -1 . The soil is a chestnut in Chinese classification and Calcisorthic Aridisol in the U.S. Soil Taxonomy classification. Soil texture of the experimental site is sandy loam with 62.75 % sand, 20.30 % silt, and 16.95 % clay (Liu et al. 2009).
In April 2005, seven blocks (107 m × 8 m) were set up in a split-plot design with 20 water and N being the two treatments. Each block was divided into two main plots receiving either ambient precipitation or additional water treatment (180 mm). Each 8 main plot was divided into six subplots which were randomly treated with four levels 10 [N 10 ], and 15 [N 15 ] g N m -2 yr -1 ), P (10 g N m -2 yr -1 ), and combined N and P addition. This study only concerned about the treatments of four levels of N addition under ambient precipitation. Nitrogen was added as urea pellets with half of it applied in early May and the other half in late June. 5

Field sampling and measurements
Soil Samples (0-10 cm) were collected from each N plot from four out of seven blocks on August 2013. For each plot, five random cores were retrieved after removing litter layer and stored at 4 º C under further analysis. Soil aggregates were isolated according to Dorodnikov et al. (2009)  Soil pH was measured in a 1:2.5 (w/v) soil-to-water slurry of soil aggregate samples with a digital pH meter. The concentration of DOC was extracted by adding 50 mL of 0.5 M potassium sulfate and analyzed by a TOC analyzer (High TOC, Elementar). Soil DIN concentration was extracted with 50 mL 2 M KCl from 10 g 20 fresh soils (McLeod, 1992) and determined colorimetrically using an AutoAnalyser III continuous Flow Analyzer (Bran & Luebbe, Norderstedt, Germany). Soil AP was 9 measured using the Olsen method (Olsen et al., 1954). Briefly, 2.5 g of the soil fraction was mixed with 50 mL 0.5 M NaHCO 3 (pH 8.5) and 5 g of phosphorus-free charcoal. The mixture was shaken at 150 rpm for 30 min and filtered. The P concentration of P in filtration was determined by molybdenum blue colorimetric method (Murphy and Riley, 1962). Soil DOC and available nutrients of control plot was listed in Table 1. The concentrations of exchangeable Ca and Mg were represented as cmol per kg soil 15 aggregate and it was mg per kg soil aggregate for available Fe and Mn.

Statistical analysis
The normality of data and homogeneity of variances were determined by Kolmogorov-Smirnov test and Levene's test, respectively. Two-way ANOVAs were executed to determine soil aggregate size (S), N addition, and their interactive effects 20 (S×N) on the ratios of DOC:DIN, DOC:AP, DIN:AP, exchangeable Ca:Mg, and available Fe:Mn. Pearson correlation analysis was used to determine the relationship 10 between soil pH and these ratios within each soil aggregate classes. All the correlation analysis was performed in SPSS 16.0 (SPSS, Inc., Chicago, IL, U.S.A.) and statistical significance was accepted at P<0.05.

N addition effects on soil available C:N:P
The ratio of DOC:DIN varied from 1.4 to 3.4 within large macroaggregates, from 1.0 to 1.5 within small macroaggregates, and from 1.0 to 1.3 within microaggregates (Fig. 1a). The DOC:DIN ratio was significantly higher in large macroaggregates than in both small macro-and microaggregates averaging across all N treatments (Fig. 1a). 10 Nitrogen addition significantly decreased DOC:DIN ratio within three soil aggregate classes (Fig. 1a). Significant interaction between N and aggregate size (S) was detected on DOC:DIN ratio (Table 2).
No difference in soil DOC:AP ratio was found among three aggregate classes (Fig.   1b). Soil DOC:AP ratio significantly decreased with N addition in both large macro- 15 and microaggregates, but it increased under N 5 in small macroaggregates (Fig. 1b).
No significant interactive N×S was detected on DOC:AP ratio ( Table 2).
The DIN:AP ratio was significantly lower in large macroaggregates as compared to both small macro-and microaggregates (Fig. 1c). Soil DIN:AP ratio showed no change along N gradients (Fig. 1c). There was a significant interaction of N×S on the 20 DIN:AP ratio (Table 2).
Nitrogen addition effects on exchangeable Ca:Mg and available Fe:Mn ratios 11 The exchangeable Ca:Mg ratio ranged from 11.0 to 15.0 across soil aggregate classes (Fig. 2a). Nitrogen addition significantly decreased Ca:Mg ratio within large and small macroaggregates (Fig. 2a). Neither aggregate size nor interactive N and aggregate size effects was detected on changes of Ca:Mg ratio.
The ratio of available Fe:Mn ranged from 0.98 to 1.57 across three soil aggregate 5 fractions (Fig. 2b). No difference in Fe:Mn ratio was detected among three soil fractions (Fig. 2b, Table 2). Nitrogen addition significantly increased soil available Fe:Mn within three soil fractions (Fig. 2b). Nitrogen addition significantly interacted with soil aggregate size to affect the ratio of available Fe:Mn (Table 2). 10 The DOC:DIN ratio was positively correlated with soil pH within three soil fractions (Fig. 3a,b,c). Significantly positive relationship was found between DOC:AP ratio and soil pH within large macro-and microaggregates (Fig. 3d,f). Significant and negative correlation between DIN:AP ratio and soil pH was only detected in small macroaggregates (Fig. 3h). Within large and small macroaggregates, the Ca:Mg ratio 15 was significantly and positively correlated with soil pH (Fig. 4a,b). Significant and negative relationships between the Fe:Mn ratio and soil pH were found within three soil fractions (Fig. 4d,e,f).

20
The stoichiometric ratios among different soil aggregates fractions Consistent with our initial hypothesis, microaggregates had lower DOC:DIN ratio 12 and higher DIN:AP ratio relative to large macroaggregates. It's generally recognized that more labile SOM concentrates in macroaggregates while more recalcitrant SOM in microaggregates (Jastrow et al., 2007). Additionally, SOM in microaggregates experiences more microbial processing cycles which is more decomposed as compared to macroaggregates (Six and Paustian, 2014). These resulted in lower C:N 5 of SOM in microaggregates. Our previous study suggested that microaggregates retained significantly higher inorganic N relative to large macroaggregates (Wang et al., 2015a). However, DOC and AP concentrations showed no difference between large macro-and microaggregates as suggested by our previous studies (Wang et al., 2015b. In this context, microaggregates had significantly lower DOC:DIN 10 ratio (Fig. 1a) but higher DIN:AP ratio (Fig. 1c) than large macroaggregates. Our results indicated that microaggregates were preferential sites for retaining DIN relative to DOC and AP as compared to large macroaggregates (Fig. 1a,c).
Microaggregates serve as an indicator of soil C storage capacity and help elucidate the capacity of soils to protect SOM and supply N and P nutrients (Du et al., 2014). 15 Higher DIN:AP ratio in microaggregates would be essential for available N supply, but it might eventually result in P deficiency under enhanced N loading rates in the long term.
Effect of N addition on DOC:DIN, DOC:AP, and DIN:AP ratios within soil aggregates 20 As expected, we observed a significant decrease of soil DOC:DIN ratio within all soil aggregate classes as affected by N addition (Fig. 1a). This should mainly be due 13 to supply of DIN which is directly transformed from urea fertilizer or indirectly derived from enhanced soil organic N mineralization and nitrification under urea addition (Chen et al., 2013) within three soil fractions. Indeed, our previous studies suggested that N addition increased DIN by up to 128.3%, 41.2%, and 45.7% (Wang et al., 2015a), while DOC only by 12.7%, 3.7%, and 12.7% within large macro-, small 5 macro-, and microaggregates, respectively (Wang et al., 2015b). As suggested by our previous work, N addition decreased soil pH from 7.3 to 5.6 in large macroaggregates, from 7.5 to 5.8 in small macroaggregates, and from 7.4 to 5.9 in microaggregates (Wang et al., 2015b). Soil pH was one of the main factors influencing DOC:DIN ratio as suggested by their positive correlations within three soil fraction (Fig. 3a,b,c). 10 Decline of soil pH was reported to be associated with accumulation of soil DIN (Aciego Pietri and Brooks, 2008), thus leading to the decrease of DOC:DIN ratio (Fig.   3a,b,c).
Soil DOC and DIN are important energy and N sources for microorganisms, and relative soil C and N availabilities (represented as DOC:DIN) make great contribution 15 to controlling microbial activity and nutrient cycling processes (Dijkstra et al., 2006;Wu et al., 2012). Significant decrease of DOC:DIN (Fig. 1a) might suggest that higher N inputs induced microbial C limitation relative to N across soil fractions. However, Wei et al. (2013) found that N addition increased both DOC and DIN but decreased microbial biomass C (MBC) concentration; and they suggested that soil microbes 20 were no longer C-limited as non-significant correlation between DOC and microbial biomass C. In their study, they failed to analyze the relationship between DOC:DIN 14 ratio and MBC which could be used to indicate relative C and N availabilities in determining microbial growth. In this study, coincident decrease of DOC:DIN ratio ( Fig. 1a) and MBC (see Wang et al., 2015b) indicated that repression of microbial growth under N addition was due to lower C availability relative to N (Treseder, 2008). 5 Our previous study suggested that N addition increased soil AP by 60.2% and 84.5% within large macro-and microaggregates, respectively .
Based on this, it was reasonable to detect significant decrease of the ratio of DOC:AP within large macro-and microaggregates (Fig. 1b). Soil acidification could enhance dissolution of phosphate from mineral-bound P pools while it was not the case for 10 DOC . Thus, production of DOC might not keep up with continuous dissolution of AP resulting in lower DOC:AP ratio under N addition (Fig.   1b). Significant positive correlations between soil pH and DOC:AP ratio (Fig. 3d,f) also confirmed higher AP dissolution than DOC supply under soil acidification. Under the scenario of enhanced ecosystem N inputs, decrease of soil aggregate DOC:AP 15 ratio would induce microbial C limitation relative to P (Treseder, 2008), and continuous release of AP from rock weathering would shift the ecosystem from N limitation to P limitation in the long term .

Effect of N addition on exchangeable Ca:Mg and available Fe:Mn ratios
Within large and small macroaggregates, significant decrease of exchangeable 20 Ca:Mg ratio supported our hypothesis. Soil acidification, as developed under increased SO 2 and NO x emissions, results in weathering and release of base cations 15 (Lucas et al., 2011). Selective weathering of certain base cations (Ca > Na > Mg > K) would result in change of base cation budgets and imbalance of metal ions in soils (Lu et al., 2014). Preferential loss of exchangeable Ca relative to Mg was the main reason of significant decrease of Ca:Mg ratio in large and small macroaggregates (Fig. 2a).
Correlation analyses also suggested coincident decline of Ca:Mg ratio and soil pH 5 indicating that more exchangeable Ca was leached than Mg during soil acidification in macroaggregates. However, unaffected Ca:Mg ratio in microaggregates declined our hypothesis that Ca:Mg ratio would decrease across three soil fractions. In this sandy grassland, microaggregates, containing more minerals and less sand would be a more favorable site for base cation binding and adsorption which was less affected by 10 N addition and sequential soil acidification comparing to macroaggregates (Gunina and Kuzyakov, 2014). Moreover, more decomposed SOM as indicated by δ 13 C (Wang et al., 2015a) would provide additionally strong binding surface for exchangeable Ca and Mg in microaggregates. Thus, significant relationships of Ca:Mg ratio with soil pH was not detected in microaggregates (Fig. 4c). 15 Soil Ca:Mg ratio affects plant Cu uptake and lower Ca:Mg might increase or decrease plant Cu uptake and toxicity depending on plant species and soil types (Lombini et al., 2003). Under enhanced N inputs, significant decrease of Ca:Mg ratio would further influence nutrient balance, i.e. Cu uptake of plants in this calcareous grassland. Researchers proposed that relative availabilities of exchangeable Ca and 20 Mg were more important that their absolute amounts (Bear et al., 1951;Albrecht, 1975;Kopittke and Menzies, 2007). And optimal Ca:Mg ratio ranges for plant growth 16 were identified for various plant species since the study from Loew (1892). In addition, different soil aggregate classes were featured by their various nutrient supply capacities (Mikha and Rice, 2004). Soil structure could modify the effect of cation ratios on plant growth (Zhang and Norton, 2002). These shed light on the essential role of studying soil exchangeable Ca:Mg ratio of different soil aggregate fractions in 5 evaluating soil fertility and plant growth.
For all soil fractions, significant increase of available Fe:Mn ratio denied our initial hypothesis. Decrease of the Fe:Mn ratio might be caused the antagonistic relationship between Fe and Mn during plant uptake (Somers and Shive, 1942). Soil acidification promoted weathering and desorption of micronutrients from soil 10 minerals which increased soil available Fe and Mn concentrations (Malhi et al., 1998).
Even though Fe availability generally limits productivity in calcareous grasslands, enhanced plant Mn uptake might retard plant Fe acquisition leading to higher available Fe retained in soil (Tanaka and Navasero, 1966;Tian et al., 2016) under soil acidification. Correspondingly, negative relationships of soil pH with Fe:Mn ratio ( Fig.   15 4d,e,f) suggested accumulation of available Fe relative to Mn under soil acidification as affected by enhanced N inputs. Consistent with our findings, Tian et al. (2015 and2016) found soil acidification enhanced Mn uptake and reduced Fe absorption by forbs which result in lower photosynthetic rates and growth and loss of forb species.
Plant Fe deficiency was reported to induce chlorosis and decrease photosynthesis in 20 calcareous grassland (Rogovska et al., 2007). Increase of soil available Fe:Mn ratio might pose threat to grassland productivity as potential aggravation of Fe deficiency 17 and occurrence of Mn toxicity to plant species (Tian et al., 2016). Soil nutrient imbalance would not only influence both quantity (yield) and quality of plants, but also cause impact on health of animals grazing on these plants (Mikha and Rice, 2004). 5

Conclusion
After 9-year N amendment, we found that microaggregates had lower DOC:DIN but higher DIN:AP ratio. The soil DOC:DIN ratio decreased with N addition within three fractions as a result of exogenous N supply. Nitrogen addition decreased DOC:AP ratio in large macro-and microaggregates which might induce microbial C limitation 10 relative to P. Nitrogen addition significantly decreased exchangeable Ca:Mg ratio within macroaggregates which purportly due to preferential leaching of exchangeable Ca under soil acidification as developed with N addition. Within three soil aggregate classes, significant increase of available Fe:Mn ratio might cause plant Fe deficiency and Mn toxicity as affected by ecosystem N enrichment. Changes in the ratios of 15 available elements among soil fractions that are responsible for different nutrient retention and cycling processes will cause nutrient imbalance, affect plant and microbial nutrient acquisition and chemical composition, and influence ecosystem productivity and health.    Significance level was set at P < 0.05. 31 Fig. 4 The relationships between soil pH and exchangeable Ca:Mg within large macroaggregates (a), small macroaggregates (b), and microaggregates (c), and between soil pH and available Fe:Mn ratio within large macroaggregates (d), small macroaggregates (e), and microaggregates (f). Significance level was set at P < 0.05.