Introduction

Reactive N (nitrogen) in soil is the primary nutrient source for vegetation growth1 and exerts a profound influence on the ecosystem’s C (carbon) cycle2. Researchers have found that N limitations affect the NPP (Net Primary Productivity) of ecosystems around the globe3,4. Together, elevated CO2 levels and climate change have exacerbated this limitation5,6. Even in N-rich tropical forests, N availability is a key regulator of C balance7. Based on modelling studies, global N limitation reduced C accumulation on land from 19 to 162 Pg C (Pg = 1015 g) between the pre-industrial period and the early 21st century4,8,9.

However, the global pattern of N limitation might be altered by N deposition10,11. With enhanced human activity (e.g., fossil fuel combustion and N fertilization), N deposition has increased from less than 1 Tg N yr−1 (Tg = 1012 g) in the 1860 s to 25 Tg N yr−1 in 200012,13 and will likely double over the next 25 years14. When N input to the ecosystem exceeds the demands of plants and microbial organisms, N saturation occurs, resulting in a series of changes in several processes, such as N mineralization, nitrification, nitrate leaching and C sequestration15. Experiments in Europe and North America have shown that if N deposition is 2.5–3.0 g m−2 yr−1, N saturation will occur16,17. In southern China, researchers have found that high levels of N deposition (3.6–3.8 g m−2 yr−1) led to N saturation in subtropical mature forests18. N saturation has been associated with anthropogenic N deposition19,20. In southern China, N deposition increased from 2.6 to 6.5 g N m−2 yr−1 along an urban-rural transect, resulting in N saturation in urban and suburban forests21.

The effects of N saturation on ecosystems are complex. Researchers have found that in N-saturated ecosystems, N addition does not increase foliar N22 which results in the reduction of plant photosynthesis18,23. In addition, N saturation can decrease C allocation to both leaves and wood24, restrict soil respiration25,26, reduce microbial biomass27 and increase N leaching28. However, some studies found that N addition continued to result in increased C allocation to aboveground biomass despite saturated soil N29. Furthermore, increasing of N leaching is the first, not last, variable to respond to N addition which is different from the statement in N saturation hypotheses20. The complexity of these N saturation effects may be due to the spatial and temporal limitations of different research approaches (e.g., long-term observational, gradient, and experimental studies)30.

Ecosystem models are suitable for assessing the effects of N addition on ecosystems on large spatial and long temporal scales. Many existing C-N (carbon-nitrogen) coupled models have been used to evaluate global C budgets4,8,31,32,33,34. Modelling has revealed that increased N deposition facilitated the absorption of an additional 0.3–1.3 Pg C yr−1 by terrestrial ecosystems in the 1990 s under elevated CO2 levels35. In the future, elevated N deposition will likely promote the uptake of an extra 0.81 Pg C yr−1 by forests11. However, the process of N saturation has not been considered in large-scale modelling studies.

Based on the available N saturation literature, we modified a process-based terrestrial ecosystem model — IBIS (the Integrated BIosphere Simulator) - to test the influence of N saturation on ecosystem photosynthesis, C allocation and litter decomposition. Historical N deposition and climate change data were used to drive model simulations. Model experiments were performed to examine the responses of ecosystems to enhanced N deposition and to evaluate the negative effects of N saturation on global C budgets under elevated CO2 conditions.

Results

The global pattern of N saturation

In our study, N deposition in excess of the N critical load is considered N saturation. The details of model modification, parameterization and determination of the spatial distribution of N critical are introduced in Methods. The spatial distribution of current N saturation is shown in Fig. 1a. Regions of N saturation are mainly located in the US, Europe, India and China, whereas other areas exhibit varying degrees of N deficits. The vegetation in saturated regions is mainly grasslands and temperate forests. N-saturated grasslands are located mainly in India and China, with small regions scattered throughout Europe. N-saturated forests are located in the eastern US and in Europe. N deposition in the saturation regions is high due to fossil fuel combustion and agricultural fertilization36.

Figure 1
figure 1

Global N-saturation regions.

(a) shows distribution of N-saturation regions; (b) shows the average N deposition of the main N saturated regions in different time periods. The maps were generated using ArcGIS 10.0 software (https://www.arcgis.com/) and SigmaPlot version 12.0, from Systat Software, Inc., San Jose California USA (https://www.systatsoftware.com).

The multi-year average N deposition for different vegetation types in saturated regions is shown in Fig. 1b. N deposition always exceeded the critical load in US temperate forests, whereas in European forests, N deposition decreased between 1970 and 2009, and current N deposition only marginally exceeds the N critical load. More temperate forests in Europe were located in N-saturated regions in the 1980 s, when N deposition was at its highest value of the past 40 years. In China and India, due to intensified human activity, rapid increases in N deposition resulted in N-saturated grasslands. In the 2000 s, N deposition far exceeded the N critical loads, indicating that the ecosystem was significantly affected by N saturation in N-saturated grassland in China and India.

Effects of N deposition on the global C budget

Several different simulation scenarios were used to evaluate the effects of N deposition on global C budgets. The details of these simulation scenarios are listed in Table 1. The historical changes of global average NPP and NEP (Net Ecosystem Productivity) were simulated by the NCC (Nitrogen CO2 Climate) scenario, and the results are shown in Figure S5. The average NPP was 51.3 Pg C yr−1 in the 1970 s and 55.9 Pg C yr−1 in the 2000 s, whereas the average NEP was 2.2 Pg C yr−1 in the 1970 s and 2.8 Pg C yr−1 in the 2000 s. Due to climate change, increasing CO2, and elevated N deposition, NPP and NEP increased by 9.0% and 27.2%, respectively, over the past 40 years (Figure S5).

Table 1 Design of the simulation experiments.

The effects of N deposition on global C budgets were evaluated by comparing the NNC (No N deposition Change) and NCC scenarios. The results show that N deposition had positive effects on the global C budget over the past few decades. Increased N deposition increased the NPP by 0.23 Pg C yr−1 on average (Fig. 2a), which accounts for approximately 0.44% of the global average NPP (Fig. 2b). N deposition promoted an increase in NEP by 0.09 Pg C yr−1 on average (Fig. 2c), which corresponds to 4.2% of the global average NEP (Fig. 2d).

Figure 2
figure 2

The differences in C budget between simulation scenarios NCC and NNC.

(a) and (b) show NPP difference and change percentage; (c) and (d) show NEP difference and change percentage; (e) and (f) show N deposition difference and change percentage.

Over the past 40 years, the total N deposition increased by 53% (Fig. 2e and f), although it did not result in significant increases in global NPP and NEP. The maximum effects of N deposition were observed in the 2000 s, with NPP increasing by 0.8% and NEP increasing by 6.0%. In general, global C assimilation is limited by N availability. However, increased N addition into ecosystems did not result in a significant increase in C assimilation. The imbalanced spatial distribution of N deposition may be the main reason underlying the contrast between terrestrial ecosystem C assimilation and rapidly increasing N deposition.

The distribution of N deposition is uneven across ecosystems. In this study, we used ecosystem biomes to examine the change in N deposition in different ecosystems. The ecosystem biomes were defined by Roy et al.37, and the global pattern of this dataset was shown by Beer et al.38. In this dataset, the cropland area was taken from the MODIS (MODerate resolution Imaging Spectroradiometer) land cover map39. As a single biome, croplands accounted for the largest fraction of the global total N deposition (28.7%) in the 2000 s. During the same period, the forest biomes together accounted for approximately 38.4% of the global total N deposition; tropical, temperate, and boreal forests accounted for 18.0%, 14.9% and 5.5%, respectively (Table S4). The increase in the rate of N deposition was significantly different among biomes (Table S4), as was the change in N deposition. N deposition increased rapidly in tropical forest, accounting for approximately 25.9% of the increase in the global total N deposition. Temperate forest biomes also increased markedly, accounting for 11.3% of the increase in global total N deposition. However, N deposition in the boreal forest biome accounted for only 0.7% of the global total increase. Compared with some forest biomes, some non-forest biomes, including cropland, grassland, shrubland and savanna, had more pronounced amounts and rates of increase in N deposition (Table S4).

The spatial distribution and increasing rate limit the fertilization effect of N deposition on the global C budget. Forest biomes are important for the global C budget, although the spatial disparity between the required N (for forest C assimilation and absorption) and increasing N deposition led to less significant increases in NPP and NEP. Increased N deposition in non-forest biomes did not contribute greatly to the global C budget.

Influence of N saturation on the global C budget

To evaluate the impact of N saturation on the global C budget, we designed a comparison experiment between the simulation scenarios NNS (No Nitrogen Saturation) and NCC scenarios. In the NNS scenario, the effects of N saturation were removed by modifying the equations for three C-N coupling modifiers, KP, K1 and KM, which were set to 1.0 when N deposition exceeded the N critical load. The results of a comparison between the NNS and NCC scenarios are shown in Fig. 3. N saturation caused the terrestrial ecosystem NPP and NEP to decline. N saturation resulted in an NPP decrease of 0.26 Pg C yr−1 and an NEP decrease of 0.03 Pg C yr−1, accounting for 0.5% and 1.3% of the annual average global total NPP and NEP, respectively, between 1970 and 2009.

Figure 3
figure 3

The impact of N saturation on the global C budget (NNS-NCC).

(a) and (b) show the N-saturation effects on NPP and NEP in N-saturated regions. (c) and (d) show the N-saturation effects of the vegetation types.

The negative effects of N saturation on the global C budget can be primarily attributed to temperate forests and grasslands. In temperate forests, NPP and NEP were reduced by 0.04 Pg C yr−1 and 0.005 Pg C yr−1, respectively. In grasslands, the NPP and NEP decreased by 0.2 Pg C yr−1 and 0.025 Pg C yr−1, respectively. Along with rapid increases in N deposition, the harm caused by N saturation was substantial in grasslands, which led to substantial decreases in NPP and NEP. These two vegetation types responded primarily to the change in N saturation over the past 40 years.

Sensitivity of the C budget to changes in N deposition

N deposition and N saturation have well-defined impacts on the global C budget. To compare the influence of N deposition to other global change factors (e.g., CO2, temperature and precipitation), five comparison simulations (i.e., the NCC, CNC (CO2 No Change), NNC, TNC (Temperate No Change) and PNC (Precipitation No Change) scenarios were performed using different driving factor settings, details are listed in Table 1. A summary of the relative contributions of the different driving factors is provided in Fig. 4.

Figure 4
figure 4

The contributions of changes in temperature, precipitation, N deposition and CO2 to the global C budget under global change.

(a) shows the contributions of driving factors to NPP; (b) shows the contributions of driving factors to NEP. (Temp is temperature, Prec is precipitation, N dep is N deposition). The dashed boxes in (a) and (b) indicate the contributions of N deposition to NPP and NEP, respectively, when N saturation effects are not considered.

Our results indicate that elevated CO2 promoted NPP and NEP. With increasing CO2 levels, NPP and NEP increased by 2.7 Pg C yr−1 (5.4% of the global total) and 1.4 Pg C yr−1 (51.0% of the global total), respectively. Rising atmospheric CO2 accelerates the photosynthetic rate by increasing intercellular CO2 and decreasing stomatal conductance, with significant effects on the global C budget40. Increases in the global temperature promote NPP but inhibit NEP, and changes in precipitation promote both NPP and NEP. However, due to their large standard deviations, the effects of temperature and precipitation on the global C budget are highly uncertain41.

Compared with the effects of temperature, precipitation and CO2, the contributions of N deposition to NPP and NEP were less significant. This does not mean that N deposition is not sensitive to the global C budget; N saturation weakens the contribution of N deposition. Without considering N saturation, the impact of N deposition on NPP was markedly greater than the impact of temperature or precipitation, and the impact of N deposition on NEP was nearly equivalent to the impact of temperature and precipitation. N saturation, in addition to the heterogeneous spatial distribution of N deposition, is the most important factor limiting the positive effects of N deposition on the global C budget.

Discussion

Ecosystem N saturation is a complex process characterized by soil mineralization and nitrification processes, nitrate leaching from water, the C:N ratio of vegetation, and other factors. In this study, the N critical load was used to determine ecosystem N saturation. The concept of the N critical load was used to establish the N input level that an ecosystem can tolerate without significant harmful effects42, which can be calculated using empirical methods or by soil process models. Compared with the methods used for site observations (i.e., N leaching or isotopic indicators), N critical loads reflect the effects of N deposition on the ecosystem at a regional scale. Furthermore, the N critical load has been used by many researchers for different vegetation types around the world, particularly in regions enhanced by N deposition. Therefore, it is reasonable to use the N critical load as an indicator of N saturation.

The N saturation regions identified in our simulations have also been identified in previous studies of regional N saturation. N-saturated regions were typically characterized by dense population and industry. N saturation was most prevalent in temperate regions with strong anthropogenic N emissions. Consistent with the simulation results of the present study, N-saturation regions in U.S. and European forests have been reported in previous studies. More than 25% of European forests are N-saturated19. In the northeastern US, results from several observation sites have shown that N saturation is a frequently occurring phenomenon22. In some East Asian forest regions, N deposition was higher than in Europe and the US, although N saturation did not occur. Some researchers believe that despite the high levels of N deposition in these forest regions, the ecosystem remains N-deficient due to the short history of elevated N deposition43.

N deposition is important to the effects of elevated CO2 on C budgets as it supplies more available N in the soil. In almost all biomes, soil mineral N declined under elevated CO2 (Fig. 5a). More rapid decreases in soil mineral N were found in boreal forest, grassland and shrubland biomes where N deposition is lower. Smaller decreases were observed in regions that are strongly affected by the increase in N deposition, such as temperate and tropical forest. If N deposition had not increased over the past few decades, soil mineral N would have declined more rapidly in temperate and tropical forests. The increased N deposition meets the increasing N demand of CO2 fertilization. Simultaneous increases in atmospheric CO2 and N deposition are significant for global C cycling because they promote increased C assimilation and absorption. However, the current effects of N saturation and the spatial distribution of N deposition have limited the contribution of N deposition to the global C budget.

Figure 5
figure 5

The change of NM (soil mineral N) in biomes and N-saturation regions.

(a) shows the average change of NM from 1960 to 2009 in biomes and N-saturation regions; (b) shows the average change of NM from 2000 to 2009 in N-saturation regions.

N saturation was also affected by increasing atmospheric CO2 concentration. Elevated CO2 affected not only global C cycles but also global N turnover. As shown in Fig. 5b, simulated soil mineral N declined, even in N-saturated regions. This decline can be mainly attributed to elevated CO2, which increases plant N demand and decreases N turnover. According to our simulations, the effect of CO2 on N turnover is larger than the effect of N deposition, although N deposition has a strong effect on N turnover in N-saturation regions (Table 2). In some CO2-enrichment experiments and field observation studies, elevated CO2 increased N uptake in the ecosystem44,45,46. From a global perspective, over the past few decades, the rising atmospheric CO2 concentration has been the main factor driving increasing global C sequestration. In such a process, increased soil mineral N is absorbed into vegetation biomass, and N retention in soils declines at elevated CO2 levels. Although ecosystems have several mechanisms to compensate for N deficiency, such as strengthening N fixation and enhancing decomposition, soil mineral N may still decline, as reported in FACE (Free-Air CO2 Enrichment) experiments44,47. Therefore, along with the soil mineral N decline resulting from elevated CO2, the ecosystem N saturation was changed.

Table 2 CO2 and N-deposition effects on N turnover (N in biomass is g N yr−1; N uptake is g N m−2 yr−1).

In the future, the negative effects of N saturation on the global C balance will continue to be complex. Trade-offs between the effects of rising N deposition and elevated CO2 fertilization will determine the fate of soil mineral N. Rapidly increasing CO2 levels increase the severity of the effects of N limitation; some regions of N saturation will become more sensitive to N limitation. However, if N deposition increases rapidly, the distribution of N saturation will be extended, and the negative effects of N saturation will be enhanced.

Our research has some limitations. A major limitation of this study involves the determination of the soil mineral N level when N saturation occurred. We used the N critical load to represent the soil mineral N when N saturation occurs in the IBIS model. However, due to the limited number of observational and experimental studies, a complete global distribution of N critical loads could not be determined. Using the average N critical loads for different vegetation types to determine N saturation contributes uncertainty to the estimation result. The Markov chain Monte Carlo method is a powerful way to determine model parameters and could also contribute uncertainty to the estimation. With the increasing availability of N-saturation observation data, the Markov chain Monte Carlo method could be used to estimate the magnitude of the model uncertainty in the future.

Methods

IBIS model description

The original IBIS model48,49 tracks soil N along with SOC (Soil Organic Carbon), although there are no soil N controls based on vegetation productivity. Liu et al.50 incorporated a largely complete N cycle module into the IBIS model, which focuses on new N feedback controls for both aboveground C assimilation and belowground SOC decomposition while imposing a balanced N budget requirement. Several new control factors were introduced into the IBIS model to control the C-N cycling process, such as KP, K1 and KM, which are modifiers of plant biomass construction, soil N immobilization and soil organic matter mineralization, respectively. More details about the N cycle module are provided in Supplemental Information SI.1.

Modification of the IBIS model

The modification to IBIS put forth by Liu et al.50 focused on the response of an ecosystem to increased soil N, and it did not account for the effect of excessive soil N. The three important N cycle control modifiers KP, K1 and KM were utilized by Liu et al.50. When NM (soil mineral N) exceeds 2 g m−2, KP, K1 and KM are fixed at 1 (Fig. 6a), indicating that the ecosystem contains sufficient N to support plant growth and will not respond to additional N input. The work of Liu et al.50 focused on Canadian boreal forests, where the soil N deficit limits forest growth. However, on a global scale, N saturation caused by human fertilization and N deposition is significant and should not be ignored. To simulate ecosystem responses to N saturation, we further modified IBIS based on recent findings to include three aspects of the N saturation effect on the C budget: C assimilation, C allocation and SOC decomposition.

Figure 6
figure 6

The conceptual N-control modifiers in IBIS.

(a) modifiers used in IBIS by Liu et al.50; (b) modifiers used in IBIS in this study. NMmax is the maximum N available in the soil, NMS is the soil mineral N when N saturation occurs, and NM is the soil mineral N.

Many studies have focused on the positive effects of N addition on C assimilation in N-limited ecosystems3,51. Only a few studies have focused on the effect of N saturation on C assimilation in long-term N fertilization experiments. The results of the N saturation experiments conducted by Magill et al.23 indicated that following 15 years of N addition, the forest NPP markedly declined. However, this decline varied among forest species. Long-term N addition resulted in a 37% decline in the NPP of Korean pine forest, whereas in hardwood forest, the NPP decline was 6% following partial treatments. Mo et al.18 found that the most extreme N treatment caused a 10% decrease in the NPP in tropical forests in southern China. Based on these studies, we parameterized the maximum N saturation effect such that the NPP decreased by an average of 17.6%.

Previous studies have indicated that N saturation typically leads to an increase in C allocation to aboveground biomass and a decrease in C allocation to belowground biomass18,25,52. Litton et al.53 reviewed forest C allocation studies worldwide and found that when challenged with excessive N input, C allocation to roots decreases by 35% on average, whereas C allocation to leaves and wood increases by 10% and 25%, respectively.

In N-limited ecosystems where N deposition ranges from 0.5 to 1.0 g m−2 yr−1, additional N input promotes the decomposition of SOC54. However, in N-saturated ecosystems, greater N input has negative effects on SOC decomposition. Janssens et al.25 used meta-analysis methods to review the responses of SOC decomposition to rich N fertilization in 36 experiments conducted in forest ecosystems.

They found that SOC decomposition declined by 15% on average due to N saturation.

To incorporate the effect of N saturation on C assimilation, C allocation and SOC decomposition, we modified the KP, K1 and KM factors in IBIS as Fig. 6b shows. The new equations used to calculate KP, K1, and KM are as follows:

where NMmax is the maximum N available in the soil, NMS is the soil mineral N when N saturation occurs, and NM is the soil mineral N.

NMS is the key parameter in equations (1), (2) and (3) and is related to the occurrence of N saturation. Many previous studies have used the N critical load to associate N deposition and ecosystem damage. Empirical N critical loads for the US, Europe and China can be found in the literature55,56,57. In this study, we used N critical loads to determine NMS in the IBIS model. These results and classified N critical loads according to the putative vegetation types modelled in IBIS are shown in Table 3. The empirical N critical loads in other parts of the world were taken as the average value of the US, China and Europe.

Table 3 N critical loads in vegetation types.

In this study, empirical N critical loads were considered the maximum N deposition level resulting in no harm to the ecosystem. If N deposition exceeds this value, N saturation will occur. When N deposition reaches the N critical load, the soil mineral N content becomes important for calculating the N cycle parameters, including KP, K1 and KM. Thus, we set the N deposition level equal to the empirical N critical loads in the ecosystems detailed in Table 3; then, we set the key parameters (KP, K1 and KM) of the N cycles to 1. This means that there is no positive or negative effect of N control on the soil mineral N content when the N input reaches the maximum ecosystem demand. Thus, the soil mineral N critical load under N deposition is taken to be NMS in the IBIS model.

Driving data and model experiments

IBIS climate inputs include historical monthly precipitation and temperature data as well as the monthly average cloud fraction, wind speed, number of wet days per month, and relative humidity. All of these data were extracted from CRU (Climate Research Unit). The surface condition data consist of the vegetation cover fraction, initial biomass C, initial soil C, soil texture, and topography. Vegetation cover fractions were calculated based on the 300 m resolution GLOBCOVER 2009 map (http://www.esa.int/due/ionia/globcover). Initial biomass was derived based on Olson’s World Ecosystem database58, and soil C was derived from the Global Organic Soil Carbon and Nitrogen datasets59. The global atmospheric CO2 concentration was derived from ESRL (Earth System Research Laboratory) measurements obtained at NOAA (National Oceanic and Atmospheric Administration) and were used as the global average CO2 level (GLOBALVIEW-CO2, 2011).

Monthly N deposition data collected between 1970 and 2009 were estimated based on the NO2 column density associated with the output of the atmosphere chemistry transfer model MOZART (Model for Ozone and Related Chemical Tracers) using a previously reported method60.

Our simulation was divided into 3 stages. The first was the pre-heat stage from 1851 to 1900, where the average monthly climate data were used to drive the model to C equilibrium. N deposition at this stage was set to a baseline value of 0.05 g m−2 yr−1. The second stage was from 1901 to 1969. Actual monthly climate data were used to drive the model. The N deposition rate of 1970 was used at this stage. The third stage was from 1970 to 2009. In this stage, both the actual monthly climate data and monthly N deposition data were used to drive the model to simulate the global C budget.

To evaluate the effects of N deposition and other factors on global C budgets, six different simulations were used. The details are listed in Table 1. In the NCC scenario, all of the driving factors were dynamic; in the other five scenarios, one of the driving factors was held constant to determine the effect of each driving factor on C cycling. The NNS scenario is used to evaluate the N saturation effects on global C budgets. In this scenario, KP, K1 and KM were set to 1 in N-saturated regions when N deposition exceeded the NMS.

Model validation

Model outputs were validated against published literature and datasets33,38,61,62. Literature results and MODIS products were used to validate IBIS GPP (Gross Primary Productivity), NPP, and NEP. Our simulated GPP (122.3 ± 3.3 Pg C yr−1) is similar to the multiple-year average GPP of MTE (Model Tree Ensembles) (119 ± 6 Pg C yr−1) during the period 1980–2009, and the distribution of biases between the IBIS GPP and MTE GPP is reasonable (details shown in Supplemental Information SI.2.1). Based on a literature review, our simulated NPP is consistent with previous findings. From 1990 to 2009, the IBIS-simulated multi-year average NPP was 53.8 Pg C yr−1; this value is similar to the values of other models that consider C-N coupling and lower than those that include only a C cycling module. Our addition of the N saturation module in IBIS improves the simulated NPP, making it similar to the MODIS NPP (details in Supplemental Information SI.2.2). A global multi-year average NEP was generated via different methods to validate our simulated NEP (Table S3). The validation results show that the simulated NEP falls within the range of previous studies. The response of the ANPP (aboveground NPP) to N addition was examined based on datasets collected from N-addition experiments in grasslands and forests (Supplemental Information SI.2.4). Our simulated ANPP increased by 18 ± 13%, which is similar to the observation results (23 ± 9%). Thus, the sensitivity of N control to the global C budget in the IBIS model is reasonable.

Additional Information

How to cite this article: Lu, X. et al. Simulated effects of nitrogen saturation on the global carbon budget using the IBIS model. Sci. Rep. 6, 39173; doi: 10.1038/srep39173 (2016).

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.