Introduction

Plant photosynthate is the primary source of soil organic C, which is essential for the cycling of soil organic C (SOC), for C sequestration and as a substrate-C source for microorganisms in the rhizosphere (Weintraub et al. 2007; Ge et al. 2012; Tian et al. 2013a). Therefore, quantification of this flux as it relates to the SOC pool will make an important contribution toward our understanding of the global C cycling process. The extent of root-derived C depends on the input from photosynthesized C. It comprises root exudates that are dominated by low molecular weight compounds such as carbohydrates, amino acids and organic acids (Nguyen 2003; Kuzyakov and Jones 2006). These compounds are important sources of dissolved organic carbon (DOC) and are readily available for soil microorganisms (Lu et al. 2002a; Ge et al. 2012). Part of the root-derived C is used for growth of the roots themselves (Farrar et al. 2003; Boddy et al. 2007), whereas another fraction is biodegraded by soil microorganisms (Kaštovská and Šantrůčková 2007). Nevertheless, below-ground C cycling remains poorly understood, due to the complexity of root exudate composition, the high degree of spatial heterogeneity in the rhizosphere, and a lack of available methods for its measurement within the rhizosphere (Walker et al. 2003; He et al. 2004; Kuzyakov and Jones 2006). Therefore, quantification of root-derived DOC and microbial biomass carbon (MBC) are essential for understanding processes such as microbial activity and soil respiration, providing further insight into C cycling in the plant-soil system.

The fate of photosynthesized C can be affected by N fertilizer application (Zagal et al. 2001; Kuzyakov et al. 2002a). However, the contribution of photosynthesized C to SOC pools varies with the N application rates, depending on experimental conditions, such as light, temperature, CO2 concentration, moisture, nutrition status, and soil texture (Zagal et al. 1993; van Ginkel et al. 2000), and on plant factors such as plant species, variety, and growth stage (Swinnen et al. 1994a, b; Kuzyakov et al. 2002b).

Upland crop plants such as maize, ryegrass, and barley are often used to determine the amount of plant-derived C entering soil and its subsequent metabolism. Kuzyakov et al. (1999) found that the total amount of 14C-labeled C in the soil plus roots ranged from 8.2 to 27.7 % of total ryegrass-assimilated C after 8 days of 14C-pulse labeling. Johansson (1992) reported that at low N levels, 32 % of the net assimilated 14C was translocated below ground, whereas at high N levels, 27 % was translocated. In the paddy ecosystem, N supply affects both rice C assimilation (as indicated by grain yield) and soil C sequestration (Cai and Qin 2006).

Rice is harvested annually from 165 million hectares of land worldwide and provides the basic food for nearly half the world’s population (FAO 2011). In addition, the paddy ecosystem can sequester more C than upland soils, slowing the increase in atmospheric CO2 concentration (Wang et al. 2007; Wu 2011). Therefore, potential changes in paddy soil C storage in response to climate change are of considerable interest (Ge et al. 2012). Hence, studying the processes of rice C distribution and transformation, and quantifying the amount of photosynthesis-derived C inputs into the soil are critical activities required to increase our understanding of global C cycling and the ecological functions of paddy ecosystems. However, there have been few direct studies of the interactions between C assimilation by photosynthesis and C cycling in rice-soil systems. Rice-derived C inputs can be partitioned into MBC- and DOC-derived sources, and N fertilizer can affect this partition of rice-derived C inputs during growth. This information will also facilitate studies of the amount and fate of plant-derived C that enters the soil; these processes are currently unclear, particularly with regard to root exudation.

We studied the effects of N fertilizer rates on the allocation and fate of assimilates by labeling rice with 14CO2 during vigorous growth (including the entire tillering stage). The objectives of this study were to (1) quantify the photosynthesized C inputs to SOC pools in the rice-soil system, and (2) evaluate the contribution of photosynthesized C to DOC and MBC at various N application rates.

Materials and methods

Soil

A typical stagnic anthrosol (Gong et al. 2009), developed from a granitic red soil, was selected from a rice field (113°19′52″E, 28°33′04″N, 80 m a.s.l.) located at Changsha Research Station for Agricultural and Environmental Monitoring. The station is in a subtropical region of China, with an annual mean temperature of 17.5 °C, rainfall of 1,300 mm, sunshine of 1,663 h, and frost-free period of 274 days. Soil samples were collected from the plow layer (0–20 cm) and sieved field moist (<4 mm) (water content, 14.8 %) to remove coarse plant residues. The soil contained 7.51 % clay, 68.36 % silt, 24.13 % sand, 18.1 g organic C kg−1 soil, and 1.80 g total N kg−1 soil, and had a pH of 5.56 (1:2.5, soil/water ratio).

Treatments

Four treatments, each with four replicates, were established as follows: No N supplied (N0), 10 mg N kg−1 soil (N10), 20 mg N kg−1 soil (N20), and 40 mg N kg−1 soil with rice growth (N40). The prepared soil (1.50 kg on an oven-dry basis) was mixed with (NH4)2SO4 to obtain the appropriate N levels, and each treatment sprayed with the same concentrations of NaH2PO4 and KCl (20 mg P kg−1 soil and 80 mg K kg−1 soil, respectively). The soils were then placed into plastic pots (20 cm diameter and 5 cm height) and sufficient deionized water was added to maintain a 1 to 2 cm depth of water covering the soil surface throughout the rice cultivation period.

14CO2 labeling of rice

The plant-growth chamber for uniform 14CO2 labeling of rice was described by Ge et al. (2012, 2013), with some minor modifications. Briefly, three 25-day-old rice seedlings (Oryza sativa L., two-line hybrid rice ‘Zhongzao 39’) were transplanted to each pot on May 29, 2012. All rice plant pots were transferred to an automatically controlled gas-tight growth chamber. The experiment spanned a 14CO2 labeling period from June 4 to July 11, 2012, in the growth chamber (area 110 cm × 250 cm, height 180 cm). The soil surface was covered by a black plastic sheet to avoid algal photosynthesis on the surface soils and the rice shoots exposed to 14CO2 for 36 days. The 14CO2 was supplied to all pots and irrigation water was introduced through a nylon tube (5 mm inner diameter) connected to each pot.

The 14CO2 was generated from the reaction between 14C-Na2CO3 (1.6 × 104 μg mL−1 and 16.5 × 103 Bq mL−1) and HCl (2 M) in plastic beakers placed inside the chamber, giving a concentration between 360 and 380 μL14CO2 L−1 (Shsen-QZD, Qingdao, China). When the CO2 concentration in the chamber fell below 360 μL L−1, more 14CO2, generated by the above reaction, was introduced into the chamber. Conversely, when the 14CO2 concentration in the chamber was higher than 380 μL L−1, a switch diverted the gas flow to pass through CO2 traps (1 M NaOH solution) to absorb excess 14CO2. Two temperature humidity sensors (SNT-96S, Qingdao, China) were installed: one inside the chamber, and another in the surrounding rice field in the open air. Two fans continuously circulated the air in the growth chamber. When the temperature/humidity sensor inside the chamber recorded a temperature 1 °C higher than the external value, a corresponding script in the data-logger triggered the relay to activate air-conditioning until the temperature inside the chamber was 1 °C lower than that outside. The entire growth chamber system was placed outdoors in order to maintain natural exposure to sunlight.

Sampling and harvesting

Plants were harvested after 36 days of labeling. The shoots were cut off at the stem base, allowing for separation of the roots, shoots, and soil. Any soil adhering to the root was removed by gentle agitation in 0.01 M CaCl2 (pH 6.2) for 1 min and any adhering soil was then thoroughly washed away under running tap water to remove soil absorbed on the surfaces. All roots, shoots, and a small soil sub-sample were dried to a constant weight in an oven at 70 °C for total C and 14C analyses. The remaining soil was stored at 4 °C until required.

Analytical methods

Soil pH was determined using a pH meter (Delta 320; Mettler-Toledo Instruments Co., Ltd., China) with a soil/water ratio of 1:2.5, and soil particle composition using a laser particle size analyzer (Mastersizer 2000; Malvern Instruments Ltd., UK). Organic C and total N were measured by dry combustion using an element analyzer (Vario MAX, Elementar Analysensysteme GmbH, Germany). Soil DOC was extracted with K2SO4 (0.5 M) according to Bolan et al. (1996), and subsequently analyzed using a total carbon analyzer (Phoenix-8000). Soil MBC was determined using fumigation-extraction (Wu et al. 1990).

The amount of 14C-labeled C in soil (14C-SOC) was measured according to Wu and O’Donnell (1997). Briefly, 1.50 g soil (<0.149 mm) was weighed into a double-necked flask containing 20 mL K2Cr2O7 (0.2 M) and 30 ml H2SO4-H3PO4 (5: 1, v:v). The mixture was digested at 165 °C for 8 min, and then pure O2 was continuously pumped into the flask to encourage digestion for a further 10 min. The evolved CO2-C was trapped in 40 mL NaOH (0.4 M). The amount of labeled CO2 (14CO2) trapped was measured by mixing 1 mL NaOH with 9 mL liquid scintillation cocktail (Beckman Coulter, Fullerton, CA, USA) and counting for 5 min using an automated liquid scintillator (LS-6500; Beckman, Germany). The amount of soil organic C (14C-SOC) that was derived from the labeled rice plants (mg C kg−1 soil) was calculated using the following formula:

$$ {}^{14}\mathrm{C}\hbox{-} \mathrm{SOC}={\mathrm{F}}_1{\mathrm{R}}_{\mathrm{s}}/{\mathrm{R}}_{\mathrm{p}}\mathrm{W} $$
(1)

where F1 is the converting factor for the counting volume (1 mL) to the volume of the trap solution (40 ml), Rs and Rp, are the specific activities (Bq L−1; blank counts subtracted) of the NaOH trap solution and Na2C14O3 (Bq mg−1 C L−1) was used to produce 14C-CO2 in the growth chamber, W, is the weight (kg) of soil on an oven-dry basis, respectively.

The amount of 14C-DOC (i.e., the amount of DOC that was 14C labeled) was analyzed in the non-fumigated soil and determined as described above. The 14C-MBC (i.e., the amount of soil microbial biomass C that was 14C labeled) was extracted from the soil and measured (Ge et al. 2012) using an automated liquid scintillator (LS-6500, Beckman, Germany). The amounts of 14C-DOC (mg C kg−1 soil) and 14C-MBC (mg C kg−1 soil) were calculated using the following formulas, respectively,

$$ {}^{14}\mathrm{C}\hbox{-} \mathrm{DOC}={\mathrm{F}}_2{\mathrm{R}}_{\mathrm{uf}}/{\mathrm{R}}_{\mathrm{p}}\mathrm{W} $$
(2)
$$ {}^{14}\mathrm{C}\hbox{-} \mathrm{MBC}={\mathrm{F}}_3\left({\mathrm{R}}_{\mathrm{f}}-{\mathrm{R}}_{\mathrm{uf}}\right)/{\mathrm{R}}_{\mathrm{p}}{\mathrm{Wk}}_{\mathrm{c}} $$
(3)

where F2 and F3 represent the converting factor for the counting volume (1 mL) to the total volume of extract (80 mL) plus soil water (in ml); Rf and Ruf, radioactivity (Bq L−1; blank counts omitted) for the extracts of the fumigated soil and unfumigated soil, respectively; Rp, is the radioactivity of 14C-Na2CO3 (Bq mg−1 C L−1) used to produce 14C-CO2 in the growth chamber; W, is the weight (kg) of soil on an oven-dry basis, and kc, the factor (0.45) converting 14C determined into the biomass 14C (Wu et al. 1990).

Statistical analysis

All data are expressed as the means of four replicates ± SD. A one-way ANOVA with a Duncan test was used to identify differences between treatments, and linear regression analysis was used to identify significant relationships among 14C-SOC, 14C-MBC, and N rate (P < 0.05, unless otherwise stated). All analyses were performed using SPSS for Windows version 14.0 software (SPSS Inc., Chicago, Illinois, USA).

Results

Shoots, root biomass, and shoot-to-root ratios under different N application rates

After 36 days of labeling the dry matter of rice shoots and root biomass had increased significantly with increasing N application rates (Fig. 1). Their maximum values were 4.09 and 2.12 g pot−1, respectively, and were obtained at a rate of 40 mg N kg−1 soil (N40 defined as the high N level below). These values are significantly higher than the minimum values (1.90 and 1.23 g pot−1, respectively) that were obtained without added N (N0). No significant differences were observed in the yield of rice roots or total biomass (P > 0.05) between the treatments of N20 (defined as the medium N level) and N40 (Fig. 1).

Fig. 1
figure 1

Amounts of rice shoot and root biomass and root-to-shoot ratios at different N levels after continuous labeling for 36 d in a closed chamber. Different lower case letters indicate significant differences (P < 0.05) among treatments. N0–N free; N10–10 mg N kg−1 soil; N20–20 mg N kg−1 soil; N40–40 mg N kg−1 soil

Full size image

There were no significant differences in the root-to-shoot ratios, which ranged from 0.52 to 0.59 in the N-fertilized soils, but this ratio was significantly higher in the N0 group compared to N10 (10 mg N kg−1 soil, defined as a low N level), N20, and N40 (Fig. 1).

The amounts of rice photosynthesis-derived C deposited into SOC (14C-SOC)

The amounts of 14C-SOC at different N application rates during rice growth ranged from 67.6 (N0) to 314.8 (N40) mg C kg−1 after labeling for 36 days (Fig. 2). The rates of N applied had a significant effect on 14C-SOC (P < 0.05), in the following order: N40>N20>N10>N0 (Fig. 2). Further, the 14C-SOC concentrations were positively correlated with the rice root biomass (n = 4, r = 0.97, P = 0.03) (Fig. 3).

Fig. 2
figure 2

The content of 14C in soil (14C-SOC) and 14C-SOC as a proportion of rice biomass C (14C-SOC/rice biomass C (%)) supplemented with different N concentrations after a 36-d continuous 14C -labeling period. Bars indicate the standard error of the mean (n = 4). See Fig. 1 for definitions of N0, N10, N20, and N40. Different lower case letters indicate significant differences (P < 0.05) between treatments

Full size image
Fig. 3
figure 3

Relationship between 14C-SOC and rice root biomass at different N levels after a 36-d continuous 14C-labeling incubation. Points represent individual N treatment means. Lines indicate linear regression between 14C-SOC and rice root biomass: 14C-SOC rice root biomass = 244.8 × (rice root biomass) - 218.7 r = 0.967, P = 0.03. Bars represent the standard errors. For data on rice root biomass and 14C-SOC see Figs. 1 and 2

Full size image

The contribution of photosynthesized carbon to SOC (14C-SOC) can be expressed as a percentage of the total rice biomass C (Fig. 2). The percentage with N40 was 19.3 %, which is significantly higher (P < 0.05) than that of N0 (8.0 %), N10 (14.1 %), or N20 (15.2 %). There was no significant difference in 14C-SOC/rice biomass C between N10 and N20, but these numbers were significantly higher than for N0 (Fig. 2).

Contribution of rice photosynthesis-derived C to MBC

After labeling for 36 days, N addition increased both MBC and the labeled MBC (14C-MBC) compared with the treatment of N0 (Fig. 4a, b). The amount of total MBC in the N10, N20, and N40 treatments increased by 35.5, 32, and 9.6 %, respectively, compared with the N0 treatment (Fig. 4b). The effects of different N application rates on 14C-MBC contents were similar to the effects on the amounts of total MBC (Fig. 4a, b), and followed the pattern: N10>N20>N40>N0. The amount of 14C-MBC at N10 was 33.79 mg C kg−1, which was significantly higher (P < 0.05) than that observed with the other treatments (Fig. 4a).

Fig. 4
figure 4

The amounts of soil 14C-MBC (a), MBC (b), 14C-DOC (c), and DOC (d) at different N levels after 36-d continuous 14C-labeling. Bars indicate the standard error of the mean (n = 4). See Fig. 1 for N0, N10, N20, and N40. Different lower case letters indicate significant differences (P < 0.05) between treatments

Full size image

Contribution of rice photosynthesis-derived C to DOC

After labeling for 36 days, the 14C incorporation into DOC (14C-DOC) was highly dependent on N concentration, and followed the series N40>N20>N10>N0 (Fig. 4c), although N20 and N40 were not significantly different. Similarly, significant differences were observed in the amounts of DOC in different N treatment (Fig. 4d). In comparison with the N0 group, the amount of DOC increased with increased addition of N. However, there was no significant difference in the DOC values between the N20 and N40 treatment groups (Fig. 4d). In addition, under the N treatments applied during rice growth, the amount of 14C incorporated into MBC (14C-MBC) differed greatly from its incorporation into DOC (14C-DOC) (Fig. 4a and c).

Distribution of 14C in SOC pools

After labeling for 36 days, 14C-SOC as a proportion of total SOC (14C-SOC/SOC) ranged from 0.37 to 1.71 % in the different N treatments during rice growth (Table 1). The value for N40 was significantly higher than that with the other treatments (Table 1).

Table 1 Distribution of 14C in SOC pools from rice-planted paddy soils at different N levels after continuous labeling for 36 days in a closed chamber
Full size table

The 14C-MBC as a proportion of total 14C-SOC (14C-MBC/14C-SOC) ranged from 6.63 % (N40) to 24.00 % (N0) after labeling for 36 days and was inversely proportional to the N concentrations supplied (Table 1). However, there was no significant difference (P > 0.05) in the proportion of 14C-MBC between N0 and N10, which in turn were both significantly higher than either N20 or N40 (P < 0.05). Soil 14C-MBC as a proportion of total MBC (14C-MBC/MBC) ranged from 6.9 to 10.6 %, and was significantly higher with N10 than with treatments N0, N20, and N40. There was no significant difference between treatments in the N20 and N40 groups (Table 1). After the labeling period, the soil 14C-DOC as a proportion of total DOC (14C-DOC/DOC) ranged from 9.7 to 11.6 %, and was not significantly different among the N treatments. Furthermore, the value for 14C-MBC/MBC was lower than that for 14C-DOC/DOC at each N concentration (Table 1).

14C-DOC as a proportion of 14C-SOC ranged from 3.92 % (N40) to 7.78 % (N0). This was significantly higher in the N0 group when compared to other N treatments (Table 1). However, there were no significant differences between the N10 and N20 groups.

Discussion

Continuous 14C-CO2 labeling

Continuous 14C-CO2 labeling was used, which provides a considerable advantage, and is a more realistic representation of C allocation than pulse labeling, which was the method used by most previous groups (Lu et al. 2002a). In the case of continuous labeling, the rice assimilated labeled CO2 constantly, over a long period, mostly between the emergence of the first leaf and the sampling time. In addition, the distribution of labeled C corresponded to the distribution of total C, as long as it was applied from first leaf emergence to harvest time (the specific 14C activity is equal in all plant parts). Therefore, continuous labeling is particularly appropriate for the estimation of the amount of total C transferred by the plants into the soil and below-ground pools during the labeling period.

The contribution of rice photosynthesis-derived C to SOC

The main sources of SOC are usually from plant residue and root exudates (Lu et al. 2002a; Ge et al. 2013). The proportion of total plant C derived from photosynthesis that is deposited in soil can reach 40 %, depending on the plant species and environmental conditions (Zagal 1994). Hütsch et al. (2002) reported that photosynthesis-derived C deposits in soil reached a maximum of 20 %, and between 64 and 86 % of this was used for soil respiration, whereas only 2–5 % was incorporated into SOC. In our experiment, 14C-SOC as a proportion of rice biomass C ranged from 8.0 to 19.3 % (Fig. 2), which is much higher than the previously reported values of 1–5 % (Lu et al. 2002b; Ge et al. 2012), and is most likely due to the continuous labeling and inclusion of N fertilization in our study. Therefore, the labeling periods and crop growth conditions should be considered when make comparisons between these studies. In general, the SOC content is higher in paddy fields than in upland soils (Wu 2011). This may indicate that “new” SOC derived from roots in paddy soil is more stable (Yuan et al. 2013; Wu et al. 2014). This is plausible, since complex formation with active iron oxide in soil may stabilize root-derived organic matter (Pan et al. 2008). This organic matter can participate in soil aggregate formation (0.25–2 mm diameter) that in turn decreases the rate of decomposition of new SOC, although this process is weaker in upland soils (Pan et al. 2008). Carter (2002) showed that SOC in coarse micro-aggregate fractions (0.02–0.25 mm diameter) contains the active C that has been recently derived from the incorporation of fresh (or less decomposed) plant residues. In addition, Tian et al. (2013b) found that flooded condition resulted in the lower rhizodeposition for rice as compared in non-flooded condition.

Effects of N application rates on the input of photosynthesized C into SOC pools

The amount of 14C-SOC in the N0 group and the corresponding proportion of rice biomass C were both significantly lower than in the N-treated groups (Fig. 2 and Table 1). This indicates that the amount of root-derived SOC is lower in N-deficient soils, perhaps because rice roots under these conditions are more effective in their competition with soil microorganisms for root exudates when compared to roots supplied with abundant N. A relatively large quantity of photosynthesized C was incorporated into SOC when N was added at 40 mg N kg−1 soil, which is generally consistent with the findings of Anderson (1988) and Liljeroth et al. (1994). The application of high N rates can enhance the deposition of photosynthesis-derived C into the soil, because it increases plant growth and stimulates rhizosphere respiration (Liljeroth et al. 1994; Chantigny et al. 1999), which may trigger the accumulation of “new” SOC derived from the roots. However, there is little information regarding the accumulation of new inputs of SOC derived from roots at the interface of the plant–soil system. We could not determine the dynamics of rhizosphere respiration during the growing season. However, previous studies have shown that the priming effects in the rhizosphere can be enhanced in high-fertilizer soils (Dijkstra et al. 2006). We therefore presume that, in the presence of high soil N, the release of C from the roots is increased, because rice growth was enhanced in the prophase of the growth stage. In the later period, more root residues remain in the soil and contribute to the root-derived SOC pools. Roots can also grow deeper when supplied with high rates of N fertilizers (Anderson 1988), and the root activities can enhance the physicochemical protection of SOC, particularly in deeper horizons (Rasse et al. 2005). In the N0 group, root growth was constrained by the lack of N. The new input of SOC derived from the roots was also lower than in the N treatments because the root exudates were in the shallow soil layer (0–10 cm) where they decompose more readily. In contrast, Johansson (1992) showed that more photosynthesis-derived C was incorporated into SOC at low N levels. This discrepancy may be attributable to differences in the experimental conditions, such as the chemical form and absolute concentration of N, the labeling periods, and crop type (Saggar et al. 1997; Gavrichkova and Kuzyakov 2008). Consequently, further work is required to reproduce our results on a wider scale.

The provision of C to roots following photosynthesis is a major contributor to soil C components, and in particular to MBC and DOC (Amiotte-Suchet et al. 2007; Ge et al. 2012). Here, the dynamics of 14C-DOC (Fig. 4) in response to increasing N concentrations were generally similar to those of the root C biomass (Fig. 1). This suggests that root-derived DOC is probably controlled by the release of organic materials originating from the roots. In rice cultivated in paddy soil, the amount of DOC is higher than that in non-cultivated soil, increasing gradually with rice growth, and is generally correlated with the root biomass (Lu et al. 2004).

The maximum 14C incorporated into MBC was 33.8 mg C kg−1 soil for the N10 group (Fig. 4), which accounted for 10.6 % of the total MBC (Table 1). This finding suggests that rice-photosynthesized C inputs into soil having an effect on the dynamics of MBC, as found by Lu et al. 2004. In our work, the application of N significantly influenced the amount of 14C-MBC and its proportion in the total MBC in the soil (Table 1 and Fig. 4). From these observations, the overall process can be described by five linked processes: (1) decomposition of organic C in native soil is stimulated at low N levels during rice growth; (2) soil microorganisms initially use C from the native soil organic C pool to a greater degree than they use root-derived C, and this stimulates their activities; (3) the population of soil microorganisms increases with increasing available C sources, which in turn maintains the decomposition of organic C in the native soil; (4) because the C sources derived from native soil cannot sustain the increased growth of soil microorganisms, about 8 % of the microorganisms use C derived from roots, whereas the rest continue to use the original soil organic C; and (5) the amount of soil MBC depends on an increased availability of root-derived C sources. Furthermore, soil 14C-MBC as a proportion of 14C-SOC (14C-MBC/14C-SOC) declined significantly with increasing N concentration during the vigorous growing season of rice (Table 1). This suggests that root-derived MBC is a major sink of photosynthesized C within new SOC pools, and is dependent on root-mediated C deposition in paddy soil under N-deficient conditions.

Conclusions

Our results show that fertilizer N enhances rice photosynthesis-derived C deposition into the SOC pools, which is then incorporated into SOC. However, the amounts of both 14C-MBC and 14C-MBC as a proportion of total MBC (14C-MBC/MBC) were significantly increased at low N levels during rice growth. This information increases our understanding of the role of below-ground biomass in the storage of SOC in flooded rice-soil systems.