**4. Discussion**

Numerous studies have explored the effects of various factors on soils δ13C and δ15N in terrestrial ecosystems on different scales [9,11,18,27,29,54]. However, it is still unclear how co-varying climatic, edaphic and biotic factors control soils δ13C and δ15N in temperate grasslands on a large scale. To answer the above question, a large-scale soil collection across temperate grasslands was carried out along a vegetation transect in Inner Mongolia. We found that biological and edaphic factors such as MBC and total N exert more effects on soil δ15N whereas climatic and edaphic factors such as MAT and total P have more impacts on soil δ13C.

Other studies have found that variations in soil δ15N values largely depend on isotopic signatures of inputs and outputs, the input–output balance, N transformation and their specific isotope effects [29]. The factors affecting the above-mentioned processes can impact soil δ13C and δ15N signatures. In the current study, soil 15N values across the transect had positive values, with the exception of two sites. Higher soil δ15N values in drier ecosystems reflects a larger loss of mineral N through strongly 15N-discriminating processes, e.g., higher gaseous N losses caused by N-cycling microbes [34,55] and increased N mineralization and nitrification [56]. Numerous studies have suggested that climatic factors control δ15N in soil [10,11,34,57]. In contrast to previous studies, the current study did not demonstrate a clear relationship between δ15N and MAT or MAP [10,11,34]. These different findings could be ascribed to two reasons: (1) The length of our transect (i.e., the scale) was much short than those in both previous studies in the same region [10,34]. (2) As an indicator that integrates many processes affecting the N cycle, controls on soil δ15N are very complicated, including various climatic, edaphic and biological factors [29,32]. Consistently with a previous study on the Chinese Loess Plateau [58], edaphic factors such as SOC, Nt and Pt strongly influenced soil δ15N across our transect. Previous studies demonstrated that soil δ15N increased or decreased with Nt contents [10,11], but we observed a significant positive logarithmic relationship between soil δ15N and Nt along the transect. Among the various factors, MBC, i.e., a biotic factor, played a more important role in controlling the soil 15N signature than climatic and edaphic factors (Table 2). This reflects that microbial processes are responsible for soil 15N dynamics across the investigated temperate grassland, supporting our first hypothesis.

Previous studies have shown that soil δ13C signature corresponds similarly to biotic factors, such as plant residue input from litterfall and rhizodeposition, including root mortality and root exudation [59,60]. Over time, dynamics of soil δ13C are therefore largely controlled by C inputs from vegetation and subsequent microbial decomposition [13,18,61]. However, we found that climate and edaphic properties exerted greater control on soil 13C in the investigated temperate grasslands (Table 2), which is consistent with previous studies demonstrating the importance of climate on soil δ13C [13,62]. Additionally, a previous study also showed that the spatial variation of soil δ13C was related to soil texture in a subtropical lowland woodland [25]. Considering that δ13C can be regarded as an indicator of SOC dynamics, our results sugges<sup>t</sup> that SOC dynamics in temperate grasslands are largely controlled by climatic and edaphic factors since MAP and Pt most dominantly affected soil 13C values. Therefore, our results confirm our second hypothesis that climatic and edaphic factors have a higher effect on soil δ13C than biological factors. This could be because water is a critical factor limiting growth of plants and microorganisms in these arid and semi-arid temperate grasslands [63,64]. Additionally, P is a key nutrient in temperate grasslands and, together with N, co-limits plant net primary production and microbial activities [65,66]. Therefore, both precipitation and Pt affect soil 13C values by altering C input and microbial decomposition. These findings indicate that climatic and edaphic factors should be taken into account in order to better understand SOC dynamics, especially focusing on their roles in the microbial decomposition of plant residues and SOC.
