Introduction

The natural nitrogen stable isotope (15N) technique has been widely used in ecological studies of N cycling1,2 because the15N natural abundance values (δ15N) of plant and soil samples are the net results of many biogeochemical processes that can cause discrimination in 15N3. It is feasible to use 15N to investigate N cycling in large areas because the information collected in a single sampling provides direct insight into past N cycling by providing a record of the net effect of N transformations. In terrestrial ecosystems, biogeochemical and physiological processes have determined variability in the δ15N of plants and soil. Though N isotope fractionation during N mineralisation is small, soil N processes, such as ammonia volatilisation, nitrification and denitrification, involve fairly large N isotope effects; the preferential leaching and gaseous loss of the lighter14N leaves the soil enriched in 15N4. The factors that affect δ15N in plants mainly include the source of N and the forms of soil N used4.Concentrations of NH4+ and NO3− in the soil affect the isotope fractionation during N uptake by plants5,6. Furthermore, mycorrhizal fungi supply depleted 15N sources to their host plants. According to a global study, mycorrhizal plants were found to be depleted by 2.0–5.9‰ relative to nonmycorrhizal plants with respect to foliage δ15N7. The use of atmospherically deposited N can also decrease the δ15N value of plants8,9.

Studies of δ15N have been conducted in many regions around the world and some characteristics of 15N distribution, as well as relationships between δ15N and environmental factors, have been found. Studies compiling worldwide data indicate that 15N is more enriched in tropical forest ecosystems than in temperate forest ecosystems10 and soil and plant δ15N are positively correlated with MAT (mean annual temperature) and negatively correlated with MAP (mean annual precipitation)7,11.

However, the data used in these studies are mainly drawn from field investigations in North America, South America, Europe, Africa and Australia. Recently, a few studies from eastern Asia have shown inconsistencies in the distributions of δ15N and their relationships with climate. For example, negative foliage δ15N data are found in the tropical rainforests of China and Malaysia12,13. Furthermore, both positive14 and negative8 relationships are found between plant δ15N and MAT and MAP in regional studies of eastern Asia. Because of the influence of the eastern Asian monsoon, the climate differs from that in Europe and North America, with apparent latitudinal gradients of temperature and precipitation are found along the North–South Transect of Eastern China (NSTEC)15. Moreover, nitrogen deposition forms a tremendous gradient along the NSTEC16. The synchronous change of spatial distribution of temperature, moisture and N deposition might give a more complicated effect on δ15N patterns in eastern China. In this study, based on a systematic investigation of seven typical forest ecosystems along the NSTEC (Fig. 1), it was our aim to 1) describe variation in δ15N values in the foliage, O horizon and mineral soil among typical forest ecosystems in China; 2) clarify patterns of δ15N distribution in the forest ecosystems over environmental gradients; 3) explore potential key factors and mechanisms contributing to the patterns. This study provides new insights into regional patterns of the natural abundance of15N isotopes in forest ecosystems in China.

Figure 1
figure 1

Vegetation regionalisation and sampling sites in eastern China.

The NSTEC is coloured and the different colours represent different vegetation types. The full names of the sampling sites are shown in Table 1. *the map is created in the support of arcgis 9.3 (ESRI).

Full size image

Results

Frequency distribution of δ15N values in foliage, O horizons and mineral soils

The relative frequency distributions of δ15N values in foliage, O horizon and mineral soil samples in natural forest ecosystems along the NSTEC are shown in Fig. 2. Significant differences in δ15N values were observed among the different components of forest ecosystems in eastern China (F = 252.68, p < 0.001). Most foliage samples had negative δ15N values, more than 80% of the values in O horizon samples were between −2‰ and 2‰ and the values of almost all mineral soil samples were between 4‰ and 8‰. The mean δ15N values in foliage, O horizon and mineral soil samples were −1.5 ± 0.2‰, −0.9 ± 0.2‰ and 5.1 ± 0.18‰, respectively; thus, in natural forest ecosystems in eastern China, live foliage biomass was depleted of 15N and mineral soil was enriched with 15N.

Figure 2
figure 2

Relative frequency of natural δ15N in the foliage, O horizon and mineral soil samples.

Full size image

Spatial distribution of δ15N

Fig. 3 shows the δ15N values of the foliage sorted according to latitude in eastern China (R2 = 0.34, p = 0.015), indicating a more regular pattern than the distribution of N concentrations. The mean values of foliage δ15N increased significantly from −4.2‰ in the DHS subtropical monsoon evergreen broadleaved forest to 0.9‰ in the LS temperate coniferous and broadleaved mixed forest. However, the mean values of foliage δ15N decreased to −2.8‰ in the boreal coniferous forest in DXAL. The δ15N values of the O horizon were slightly greater than the values of the foliage, increasing from −3.3‰ to 2.6‰. Like the foliage, the δ15N values in the O horizon increased significantly with increasing latitude (R2 = 0.30, p = 0.043). However, the statistical significance was weaker between δ15N values in the mineral soil and latitude.

Figure 3
figure 3

N concentrations and δ15N values of the foliage in typical forest ecosystems in eastern China (sampling sites are arranged such that latitude increases from left to right).

Full size image

Temperate forest ecosystems were significantly more 15N-enriched than tropical forest ecosystems (Fig. 4). The mean δ15N value for tropical forest foliage was −2.3‰, which was 1.6‰ lower than that of temperate forest foliage. The range of δ15N values was larger in the organic layer than in the soil mineral layer and δ15N was 1.7‰ and 0.7‰ in O horizon and mineral soil, respectively.

Figure 4
figure 4

Differences in δ15N between temperate and tropical forest ecosystems (statistically significant differences were set as p < 0.05).

Full size image

The δ15N values increased with soil depth and they increased sharply from the O horizon to the mineral soil. The δ15 N values in the O horizon and mineral soil were distinctly different (F = 363.07, p < 0.001). The range of δ15N values among different sites decreased with increasing soil depth, from 3.9‰ in the top sampled layer to 2.2‰ at 40 cm below the ground surface (Fig. 5a). Moreover, the mean N concentration decreased from the upper O horizon to the lower mineral soil horizons (Fig. 5b). Similar to the variation in δ15N values, differences in N concentrations between the O horizon and mineral soil were significant (F = 245.31, p < 0.001).

Figure 5
figure 5

Variations in δ15N values and N concentrations from the O horizon to a 40 cm mineral soil depth at the different sampling sites.

Full size image

Effect of environmental variations on δ15N

There was clear latitudinal zonality in the distribution of environmental factors in eastern China, such as the MAT, MAP and N deposition (Table 1). It was clear that environmental variation exerts an obvious effect on foliage δ15N values (Fig. 6). However, these effects decreased as the depth increased because the range of soil δ15N values across sites diminished with increasing depth (Fig. 5). Foliage δ15N values were more dramatically influenced by nitrogen deposition than by climate factors. As analysed by step regression, N deposition was the key environmental variation determining the15N distribution in eastern China (F = 9.775, p = 0.007).

Table 1 Characteristics of the sampling sites
Full size table
Figure 6
figure 6

Relationships between foliage δ15N values and environmental factors (temperature, precipitation and N deposition).

Full size image

Foliage δ15N values increased dramatically with increasing N concentrations in eastern China (Fig. 7a). Similarly, N concentrations and δ15N values were positively correlated with the O horizon, although this relationship was not statistically significant. In contrast, δ15N values in the mineral soil exhibited a negative relationship with N concentrations (R2 = 0.199, p = 0.043). However, we detected no pattern in δ15N differences between the upper (0–10 cm) and lower soil layers (10–40 cm) of the mineral soil when plotted against the variation of the mineral soil N concentrations. Thus, the relationship between δ15N and N concentrations in the soil layers was driven by the soil depth.

Figure 7
figure 7

Relationships between foliage δ15N values, N concentrations and net N mineralization.

Full size image

Foliage δ15N values decreased significantly with the net soil N mineralisation rate in eastern China (no data in DGS) (Fig. 7b). However, there was no significant correlation between δ15N values and the net soil N mineralisation rate in the O horizon and mineral soil layers in forest ecosystems in eastern China.

Discussion

A meta-analysis study has indicated that the δ15N values for plant tissues and bulk soils are generally between −10‰ and +15‰ in forest ecosystems17. δ15N values of mature forest ecosystems in eastern China occupied the middle of this range (−4.2‰ to 8.4‰). The values of foliage δ15N ranged from −4.2‰ to +0.9‰, which is less than the average value of foliage δ15N from worldwide studies7,11. In agreement with broadly observed patterns18,19, foliage and O horizon samples from mature forest ecosystems in eastern China were significantly depleted in 15N relative to mineral soils (Fig. 2) and δ15N values increased with soil depth (Fig. 5a). δ15N values in soil surface samples are lower than further down in the soil because litterfall imports 15N-depleted N onto the soil surface. Another possible explanation of this pattern is that plants discriminate against 15N during N uptake19. In addition, the enrichment of 15N in deeper soils may be the result of discrimination against 15N during mineralisation and the loss of 15N-depleted nitrogen from soils due to root uptake, nitrate leaching and denitrification4,18.

δ15N values increased with increasing latitude along the NSETC in eastern China (Fig. 3). However, in contrast to the meta-analysis by Martinelli et al.10, temperate forest ecosystems were significantly more15N-enriched than tropical forest ecosystems (Fig. 4). The average value of foliage δ15N in the subtropical forest ecosystems was negative and approximately 6‰ lower than the summarised worldwide value, but it was similar to the value of temperate forest ecosystems in the worldwide synthesis10. The average value of foliage δ15N was −0.7‰ in the temperate forest ecosystems in east China, which was 2.1‰ higher than the average value from the worldwide data synthesis. The investigation in the Hainan islands, at the south end of the NSTEC, demonstrated that the foliage δ15N there is between −2.8‰ and −2.0‰13, similar to the values in our study. Nevertheless, in the tropical forest ecosystem in Malaysia, the foliage δ15N shows an even lower negative value12. Our results show that the foliage δ15N values varied significantly with climate in eastern China (Fig. 6). In contrast to the global results indicating that foliage δ15N values increase with decreasing MAP and increasing MAT7,11, foliage δ15N values were negatively correlated with both MAP and MAT in forest ecosystems along the transect in eastern China (Fig. 6). We propose several potential mechanisms to explain the δ15N patterns along the NSTEC.

First, the foliage δ15N values may be controlled by the available inorganic N sources and the ratio of NO3− to NH4+ in soils in eastern China. Higher temperatures promote microbial activity and enhance both mineralisation and nitrification rates. Net N mineralisation rates are generally higher in subtropical and tropical forest soils than in temperate forest soils and nearly all nitrogen that is mineralised in these systems is immediately nitrified. For example, soil NH4+ accounts for more than 88% of the total inorganic N in the forest ecosystem at CBS20, while nitrification represents close to 80% of the net mineralisation in subtropical monsoon forest ecosystems at DHS21. Nitrification is associated with fairly large isotope effects22 and some studies have reported that15N in NH4+ is more enriched than that in NO3− in soil solutions20,23.15N enriched NH4+ is available for absorption by plants before it has been nitrified in N-limited conditions24 in the north of the NSTEC. The15N-depleted NO3− would also be used by plants rather than left from the ecosystem in the south of the NSTEC because the N cycle is in stage 1 according to the concept of N saturation25. Therefore, the foliage δ15N increased from south to north along the NSTEC and was negatively correlated with the increasing MAT (Fig. 6) and net soil N mineralisation (Fig. 7).

Second, in recent decades, the biogeochemical cycling of nitrogen has been fundamentally changed by atmospheric N deposition as a result of fossil fuel consumption, the emission of industrial waste gases, excessive fertiliser application and rapid development of animal husbandry26. Fossil fuel emissions have depleted15N in the nitrate deposition over the past decade, as revealed by the δ15N values in a 100-m-long ice core27. Because ammonia volatilisation is a highly fractionating process and produces much15N-depleted NH327, the δ15N values of deposited ammonium are usually negative. For instant, Xiao & Liu28 reported that the mean δ15N values of nitrate and ammonium in wet deposition at seven different sites around the world (including the Chinese city Guiyang) were −1.9‰ and −5.0‰, respectively. Studies of five sites in the North China Plains found that the mean values of nitrate and ammonium in rainfall were −2.5‰ and −1.2‰, respectively29. In Guangzhou, the provincial capital of Guangdong, the annual values of nitrate and ammonium in wet deposition were −0.7‰ and −7.3‰, respectively30. The moss δ15N, which was assumed to be an integrator of the isotopic signatures of atmospheric N sources, also indicated that the δ15N of atmospheric N in the urban areas, rural areas and forests of southern China were, on average −7.5‰, −3.4‰ and −0.8‰, respectively31.

Given that the N deposition increased more than threefold from north to south in eastern China and that ammonium is the dominant form of N deposition in eastern China16, it appears that the 15N-depleted N input in tropical forest ecosystems is higher than that in temperate forest ecosystems. Plants grown in areas of heavier atmospheric N deposition have much lower δ15N values, likely as a result of absorbing more 15N-depleted inorganic N sources through their roots or stomata9,25. In our study, the influence of N deposition on foliage δ15N patterns is reflected by the significant negative relationship between the foliage δ15N values and atmospheric nitrogen deposition (Fig. 6). This negative relationship has also been observed in a forest ecosystem along an urban-rural gradient32. The intensified N deposition is also thought to be the reason why the relationship between δ15N and MAT along a grass transect in eastern China is inconsistent with the global pattern8. The difference in atmospheric N deposition is likely to be a crucial contributor to the pattern of foliage 15N signatures in eastern China. δ15N data from tree rings show a significant decline from 2.1‰ to −1.5‰ over the past 60 years, giving further evidence that N deposition decreased δ15N patterns along with the available N25. However, in contrast to Europe and America33,34,35, the history of ecosystem in eastern China suffering anthropogenically elevated N deposition is short, for the reason that enormous development of the economy began in 1980s in China. The high N deposition has not enriched 15N by increasing N losses in N-rich sites. Therefore, the impact of N deposition flux to the 15N natural abundance remains negative in the natural ecosystems in China.

Third, the existence of mycorrhizal fungi may have distorted the vertical pattern in the surface soil and depleted foliage15N at DXAL (Fig. 3, 5). The growth of trees and shrubs is often strongly N-limited because soils typically contain large amounts of organic N and low concentrations of inorganic N36,37. One way for plants to relive this N limitation is to take up amino acids from the soil through symbiotic fungi18,38. The N transfer from mycorrhizal fungi to their hosts is15N-depleted because of fractionation during metabolic processes and selective retention of specific N compounds by the fungus, as evidenced by fungal N enrichment of 3–11‰ relative to the host plant N4,19. The existence of mycorrhizal fungi could explain the low foliage δ15N and the distorted vertical pattern in the surface soil at DXAL. The presence of mycorrhizal fungi was shown to be one of the most important reasons for the enrichment of foliage δ15N from south to north along a transect in forest ecosystems in Finland3. Analyses based on worldwide data also indicate that plants infected by mycorrhizal fungi are depleted in foliage δ15N by almost 6‰ relative to non-mycorrhizal plants7.

Finally, the root distribution in the soil profiles according to different site conditions might contribute to the differences in foliage δ15N. Trees with shallower roots might absorb more 15N-depleted N because soil δ15N increases with the soil depth (Fig. 5). Moreover, differences in root distributions might be caused by variation in edaphic conditions, regardless of the plant species. For example, the root soil is only 20 to 30 cm deep at DXAL due to the existence of a permanent frozen layer, while soft soil, which is good for the growth and development of plant roots, is more than 100 cm deep at CBS. Differences due to plant species appear to be small in eastern China. The range of foliage δ15N was 5.1‰ in our study, while the range of foliage δ15N for a single tree species was also 5.0‰ in a narrower region in eastern China14.

Methods

Site description

The north-south transect of eastern China (NSTEC) extends from Hainan island to China's northern border, ranging from 108°E to 118°E at latitudes less than 40°N and from 118°E to 128°E at latitudes equal to or greater than 40°N. The transect includes 25 provinces and covers nearly one third of the land area in China (Fig. 1). A vegetation sequence is distributed along the NSTEC, which includes cold-temperate coniferous forest, temperate mixed forests, warm-temperate deciduous broadleaved forest, subtropical evergreen coniferous forest, evergreen broadleaved forest and tropical rainforest from north to south. The NSTEC is driven mainly by thermal conditions, with gradients of precipitation, atmospheric nitrogen deposition and land use intensity and it provides an ideal platform to research carbon, nitrogen and water cycles of forest ecosystems in East Asia's monsoon region39. Along the NSTEC, we chose seven forests to characterise δ15N values of different components of China's typical forest ecosystems, which included Dinghushan (DHS), Huitong (HT), Dagangshan (DGS), Donglingshan (DLS), Changbaishan (CBS), Liangshui (LS) and Daxinganling (DXAL) (Fig. 1). The specific characteristics of the sampling sites are described in Table 1.

Sample collection

From July to August 2008, foliage, litter and mineral soil samples were collected in the seven representative forest ecosystems from north to south along the NSTEC. To eliminate any systematic deviation, only non-leguminous predominant species were selected as samples to characterise the representative forest ecosystems. Three mature individual trees of each dominant species with mid-range diameter at breast height were randomly selected in each of the forest stands for collecting foliage samples. Each tree was open-grown, with full southern sun exposure. Fully mature leaves were sampled from the tips of south-facing branches in the lower crown of each tree by a tall tree trimmer. Foliage samples were collected from three individuals of each plant species and then combined into one sample.

Three sets of litter samples, comprising the Oi horizon and the Oea horizon, were randomly sampled with a trowel within a 5 cm radius. In addition, three sets of mineral soil profile samples (5–6 m apart) were collected from three depths (0–10, 10–20 and 20–40 cm) with a 5 cm diameter auger. To reduce the impact of spatial heterogeneity, the soil at each depth was collected from three cores (holes, approximately 0.5–1.0 m apart) using a soil corer for each set of samples.

Foliage and litter samples were dried to constant weight at 65°C. Mineral soils were air dried at room temperature and then sieved through a 2-mm sieve to remove roots, gravel and stones. All samples were ground into a fine powder with a planetary mill and stored in glassware. Plant and soil samples were oven dried at 65°C for 24 h before analysis. In total, we obtained 48 foliage samples from 15 species, 42 litter samples and 63 soil samples from the seven natural forest ecosystems.

N concentrations and δ15N analysis

Nitrogen concentrations and δ15N values were determined simultaneously with an automatic, online elemental analyser (Flash EA1112, ThermoFinnigan, Milan, Italy) coupled to an isotope ratio mass spectrometer (IRMS) (Finnigan MAT 253, Thermo Electron, Bremen, Germany). The standard deviation of 10 repeated samples was 0.396‰. The results of the IRMS measurement are given in δ notation. δ values of isotopes of 15N were expressed in units of per mil (‰) and calculated as follows,

Where Rsample and Rstandard are the nitrogen isotopic ratios (15N/14N) of the samples and standard, respectively. A positive or negative δ15N value indicates the enrichment or depletion, respectively, of the heavy isotope relative to the light isotope, with respect to atmospheric N2.

Statistical analysis

The sample differences of δ15N values and N concentrations were tested with analysis of variance (ANOVA). Comparison of the means was performed using Tukey's HSD test. The relationships between foliage or soil δ15N and environmental or geographical variables were evaluated using Pearson correlation coefficient. Step regression analysis was used to select the key environmental factors contributing to variability in foliage δ15N. All analyses were conducted by SPSS software package (SPSS for Windows, Version 13.0, Chicago, IL, USA). Statistically significant differences were determined by p < 0.05 unless stated otherwise.