On the fate of anthropogenic nitrogen
See allHide authors and affiliations
-
Contributed by William H. Schlesinger, October 16, 2008 (received for review September 7, 2008)

Abstract
This article provides a synthesis of literature values to trace the fate of 150 Tg/yr anthropogenic nitrogen applied by humans to the Earth's land surface. Approximately 9 TgN/yr may be accumulating in the terrestrial biosphere in pools with residence times of ten to several hundred years. Enhanced fluvial transport of nitrogen in rivers and percolation to groundwater accounts for ≈35 and 15 TgN/yr, respectively. Greater denitrification in terrestrial soils and wetlands may account for the loss of ≈17 TgN/yr from the land surface, calculated by a compilation of data on the fraction of N2O emitted to the atmosphere and the current global rise of this gas in the atmosphere. A recent estimate of atmospheric transport of reactive nitrogen from land to sea (NOx and NHx) accounts for 48 TgN/yr. The total of these enhanced sinks, 124 TgN/yr, is less than the human-enhanced inputs to the land surface, indicating areas of needed additional attention to global nitrogen biogeochemistry. Policy makers should focus on increasing nitrogen-use efficiency in fertilization, reducing transport of reactive N to rivers and groundwater, and maximizing denitrification to its N2 endproduct.
Numerous papers have documented the current human impacts to the global nitrogen cycle; humans have approximately doubled the input of available nitrogen to the Earth's land surface, largely through the industrial production of nitrogen fertilizer (1–3). Some environmental effects are obvious, including the expanding summertime hypoxic zone in the Gulf of Mexico (4) and rising concentrations of nitrous oxide in Earth's atmosphere (5). Other consequences, such as the health impacts of nitrate in drinking water are less clear (6).
This paper provides some composite estimates of the fate of nitrogen applied to the Earth's land surface, focusing on four sinks—the biosphere, groundwater, fluvial transport to the sea, and denitrification. An additional movement of terrestrial nitrogen—atmospheric transport of NHx, NOy, and other forms to the oceans—is also considered using recent literature values. The residence time of reactive nitrogen in vegetation and soils can vary over a wide range of values, depending on vegetation and its age, affecting the permanence of the biospheric sink. None of these fluxes can be estimated precisely; the goal is to provide preliminary estimates of their relative role, which can indicate potential policy solutions to the global nitrogen problem.
Providing a balanced global nitrogen budget is not easy: the element has numerous valence states and gaseous forms that escape notice and travel long distances in the atmosphere. We have little understanding of the fate of the applied nitrogen, except to note that mass-balance studies of agricultural fields indicate that a lot of it must escape from the local point of application (7, 8).
The Biosphere.
Nitrogen limits net primary production over much of the Earth's land surface (9), and to a large extent, the human impact on the global nitrogen cycle stems from our attempt to alleviate nitrogen deficiencies in agriculture by the application of fertilizer (10). Globally, ≈10% of applied nitrogen is contained in food (11); most nitrogen remaining is lost to the environment during food production and after human ingestion. There is some long-term retention of nitrogen applied to agricultural soils (8, 12–14), but much is lost to runoff, to situ denitrification, and to gaseous forms of nitrogen (e.g., NH3 and NOx) that are deposited downwind (15). Combined with emissions of NOx from fossil fuel combustion, the latter produce a chronic nitrogen enrichment of natural ecosystems, with documented effects on forests (16), deserts (17), grasslands (18, 19), coastal ecosystems (20, 21), and oceans (22).
Anthropogenic nitrogen indirectly delivered to forests can accelerate their growth, resulting in greater biomass (23, 24). The results from fertilizer trials in forestry are instructive, indicating that ≈25–30% of the applied nitrogen is retained in biomass (25) and a similar amount retained in forest soils (Table 1). Generally, coniferous forests retain more nitrogen than deciduous forests. Anthropogenic nitrogen deposited from the atmosphere may allow for the apparent acceleration of forest growth in response to rising carbon dioxide concentrations in Earth's atmosphere (41) and explain the limited evidence for progressive nitrogen limitation in CO2-enrichment studies (42). Humans have increased the atmospheric deposition of nitrogen on land by ≈46 TgN/yr (43), with about one-third of that deposited in forests (18 TgN/yr) (44), where trees provide a potential long-term sink for nitrogen in biomass. If ≈50% of nitrogen deposition is stored within the ecosystem, then ≈9 TgN/yr might be sequestered in terrestrial biomass until harvest or natural mortality. Some of this nitrogen is returned to the atmosphere as NOx or N2 when woody biomass is burned (pyrodenitrification) (45) and eventually nitrogen is released from decomposition, but a net sink for nitrogen on land is consistent with the apparent sink for carbon in terrestrial biomass. Goodale et al. (46) suggest that the retention of atmospheric nitrogen by forests in the eastern U.S. may be as high as 73%, consistent with the current sink for carbon—0.3 to 0.58 PgC/yr—in U.S. forests, many of which are recovering from agricultural land abandonment a few decades ago (47).
Fate of exogenous nitrogen applied to forest ecosystems, including studies that recorded changes in plant, litter, and soil organic matter.
Runoff and Fluvial Losses.
A large number of studies have examined nitrogen losses in surface runoff. Most of these focus on dissolved NH4+ and NO3−, but forms of dissolved organic nitrogen increasingly are recognized to contribute to the total (48). Compiling data from watersheds in the northeastern United States, Van Breemen et al. (49) found that rivers transported ≈23% of the nitrogen applied to their watersheds (cf. 26%; 15% for nitrate only) (50, 51). Recent work by Schaefer and Alber (52) suggests that transport from watersheds in warm temperate regions may be lower than in colder regions, but globally, if 23% of applied N (150 TgN/yr) is lost to riverflow, I estimate that the flux in rivers has increased from a preindustrial value of 27 TgN/yr (43) to a current value of 61.5 TgN/yr, for a net sink for anthropogenic N of 34.5 TgN/yr. Independent estimates of the recent anthropogenic riverine flux (19 TgN/yr) (53) and the total transport of N in the world's rivers, 59 TgN/yr (54) and 54 TgN/yr (55), are roughly compatible with these estimates. Much of the reactive N in rivers is denitrified in the coastal zone, returning N2 to the atmosphere (56). As fertilizer use expands globally, increases in the fluvial transport of nitrogen are likely to occur worldwide (57).
Groundwater.
Public health agencies have long recognized that excessive nitrate in water leads to the sometimes fatal condition in infants known as methemoglobinemia (58) and that regions of intensive agriculture and fertilizer use often have high concentrations of NO3− in groundwater (59, 60). Inasmuch as the nitrate concentrations in groundwater in many remote areas show near-zero values, the high nitrate concentrations found in agricultural areas are likely to be of anthropogenic origin. Compiling the results of several surveys, Spalding and Exner (61) found nitrate-N > 3 mg/L in 9.4% to 23.6% of wells in domestic areas of the rural United States. Similarly, Nolan et al. (62) compiled nitrate concentrations reported from 1,280 wells across the continental United States, calculating a median value of 8.3 mgNO3 per liter (1.9 mgN/L; cf. 3.4 mgNO3/L; ref. 63). The field sampling in these studies may be biased toward problematic regions, but the value of 1.9 mgN/L can be used to provide a lower-limit estimate of the anthropogenic impact on the annual flux of nitrogen to groundwater, if we assume that it might otherwise be nitrogen-free.
Groundwater flux (i.e., the new inputs to the saturated zone) is estimated at 1016 liters annually (64), ≈10% of global terrestrial precipitation and roughly compatible with the traditional assumption that the base flow of rivers carries ≈10% of precipitation to the sea (65). If we assume that the groundwater flux in North America is proportional to the global surface runoff generated there (14%) (66), then the input of nitrate to groundwater in North America is ≈3 TgN/yr (1.9 mgN/L × 0.15 × 1016L/yr; equivalent to ≈1.15 kgN/ha/yr for North America). This analysis does not consider additional nitrate that is transported to and denitrified in the saturated zone, which is included (44 TgN/yr) as part of the global estimate of denitrification (see below; 56). Anthropogenic inputs of nitrogen in North America are ≈20% of the world's total (67), so the global flux of nitrogen to groundwater may be as large as 15 TgN/yr. Because the mean residence time of groundwater is often thousands of years (68), it offers a long-term sink for nitrogen. Among comparative estimates, Lin et al. (69) developed a model to predict nitrate leaching of 12 TgN/yr to groundwater, adding to a preindustrial background flux of 14 TgN/yr. Van Drecht et al. (55) suggest that 10% (≈5 TgN/yr) of the nitrogen transport in rivers stems from groundwater, where N is largely derived from historical fertilizer use.
Some recent studies report large nitrogen accumulations in the vadose zone of desert ecosystems, totaling ≈3,000 to 15,000 TgN globally (70, 71). This nitrate has accumulated over an unknown period, so the pool is difficult to relate and probably not relevant to the fate of a large fraction of the anthropogenic nitrogen emitted during the past 100 years. The rate of accumulation of this nitrate in deserts (≈1 kgN/ha/yr) (72) may indicate a lower limit for the transport of nitrate to groundwater, with higher values expected in more mesic regions.
Denitrification.
Environmental scientists are increasingly realizing the complexity of microbial transformations of nitrogen in soils and sediments (73, 74). Whereas many early textbooks of microbiology recognized nitrate and N2 gas as the sole products of nitrification and denitrification, respectively, we now realize that a small percentage of the nitrogen passing through these biochemical pathways is released as NO and N2O (75). N2O is of particular interest, because it is a powerful greenhouse gas in Earth's atmosphere and destroys stratospheric ozone. Studies of the stable isotopes of nitrogen have sharpened our perceptions of nitrogen biogeochemistry; accumulations of the heavier isotope of nitrogen (15N) in soils (76) suggest that denitrification is more widespread and of greater significance than we realized just a few years ago. For one wet tropical forest, Houlton et al. (77) attribute 24 to 53% of ecosystem losses of nitrogen to denitrification. Losses of nitrogen to the atmosphere via denitrification on the land surface result in a global mean residence time for nitrogen on land of ≈100 years (68).
Seitzinger et al. (56) compiled data for denitrification from a variety of ecosystems, estimating the current global rate of 573 TgN/yr, with 22% (124 TgN/yr) stemming from the land surface and 110 TgN/yr from wetlands. These high values are somewhat difficult to reconcile with estimates nitrogen inputs to the land surface, especially if we assume that the nitrogen cycle was in steady-state in the preindustrial era. For comparison, Galloway et al. (43) suggest global denitrification of 389 TgN/yr, with ≈100 TgN/yr derived from upland and freshwater environments. Some of this flux is natural, but at least a portion is stimulated by anthropogenic additions of nitrogen fertilizer to the land surface, thus representing a sink for applied nitrogen. Globally, N2O comprises ≈2.6–3.9% of the denitrification flux, calculated by using the global N2O flux from soils, wetlands, and the sea surface (5) and global denitrification from Seitzinger et al. (56) or Galloway et al. (43), respectively.
Nitrous oxide (N2O) is a byproduct of nitrification and denitrification and is increasing in Earth's atmosphere by ≈0.3%/yr (5). The approach to estimating recent changes in global denitrification is to divide the observed increase in N2O in Earth's atmosphere (≈4 TgN/yr) by an estimate of the ratio of N2O to the total of N2+N2O produced by soil microbes. Table 2 (see also Table S1)includes values for the N2O to (N2+ N2O) ratio in the efflux from a variety of ecosystems under natural and N-fertilized conditions. As noted in many earlier reports (78), there is considerable range in these values. For wetlands, including stream and lake sediments, the values are always low; N2 dominates the efflux in anerobic environments (79–81). For upland ecosystems, a significant percentage of the efflux occurs as N2O, with an unknown proportion of this derived from nitrification rather than denitrification. The mean ratios N2O/(N2 + N2O) for upland, agricultural, and wetland soils are 0.49, 0.37, and 0.082, respectively. To provide an estimate of the human impact on terrestrial denitrification, assume that the rise in N2O in Earth's atmosphere is solely from denitrification on land. If 53% of denitrification occurs in the upland soils, dominated by agriculture, with a ratio of 0.37, and 47% occurs in wetlands with a ratio of 0.08 (57), then the weighted mean ratio is 0.23, and the calculated total rate of denitrification is now 17 TgN/yr greater than in preindustrial times. Any contributions of N2O from nitrification would reduce this value, because it would not be associated with the production of N2. An identical estimate for the human-induced change in terrestrial denitrification is given by Galloway et al. (43).
Mean N2O-yield values from various laboratory and field studies of denitrification
Seitzinger (82) compiled data showing a wide range of values for the N2O/(N2+N2O) mole fraction in the gaseous efflux from coastal marine ecosystems. Most values were very low, but see ref. 83. Near-shore denitrification may be as large as 79 TgN/yr (56), of which 21 TgN/yr may derive from human distribution of available nitrogen (43). Thus, a substantial fraction of the nitrogen leaving the land surface is likely to be denitrified in coastal waters. In this analysis, changes in pelagic denitrification are assumed to be minor (43), potentially with a significant fraction contributed by the anammox reaction that produces no N2O (73).
Xu et al. (84) report significant correlations in the flux of CO2 and N2O from soils (cf. ref. 85). Soil denitrification and N2O flux may increase globally as future, rising atmospheric CO2 stimulates plant growth, accumulation of carbon in soils, and greater denitrification. Kammann et al. (86) report large increases in the loss of N2O from a grassland grown under Free Air CO2 Enrichment (FACE) conditions. In contrast, Phillips et al. (87) found only small changes in the flux of N2O from soils in a forest grown at ambient +200 μl/L atmospheric CO2.
Atmospheric Transport of Available Nitrogen to the Marine Environment.
Atmospheric transport of reduced and oxidized nitrogen trace gases and various forms of organic nitrogen to marine ecosystems represents an additional sink for anthropogenic nitrogen applied to the land surface. Duce et al. (22) indicate that the flux of these constituents to the ocean surface has increased by 48 TgN/yr from preindustrial times to 2000, representing a net sink for anthropogenic nitrogen from the land surface. Nearly 70% (33 TgN/yr) is derived from inorganic forms—NHx and NOy—and the rest is organic N. For the 1990s, comparable values for the inorganic portion are 25 TgN/yr (43) and 39 TgN/yr (88). Of course, the atmospheric transport of reactive nitrogen from land to sea via the atmosphere is recognized as a major source of nitrogen causing environmental degradation in coastal waters and perhaps also in the open ocean (22).
Conclusions
The sum of these preliminary, generally upper-limit estimates of sinks for anthropogenic nitrogen applied to land is ≈124 TgN/yr—slightly less than the current estimate of the industrial production of available nitrogen by humans (150 TgN/yr; 125 Tg from industrial production plus 25 Tg from fossil fuel combustion; Table 3). Thus, the budget for the terrestrial portion of the modern nitrogen cycle is nearly balanced, although a number of the “sinks” (e.g., surface and groundwater pollution) are associated with serious environmental problems. Further work will be needed to refine these estimates and identify other sinks for anthropogenic nitrogen applied to the land surface. Studies of nitrogen retention in fertilized agricultural soils (12) and of submarine discharges of nitrogen to coastal marine environments (83, 90, 91) are worthy of examination.
Budgets for nitrogen on the global land surface
The comparative magnitude of these fluxes gives some indication of where better environmental management might reduce human impacts on natural and managed ecosystems. All efforts to minimize human impacts on the global nitrogen cycle will benefit from improved efficiency in the application of nitrogen fertilizer (92). With the era of cheap energy now ending, economics alone may dictate a greater efficiency of fertilizer use. However, policy makers should consider future regulations on losses of nitrogen to runoff and perhaps even to institute a cap-and-trade system to reduce wasteful inputs to the global nitrogen cycle.
The loss of applied nitrogen in rivers causes substantial disruption to estuarine and coastal marine ecosystems (93). To minimize such transport, land-use planning should encourage the maintenance and expansion of wetland ecosystems that convert nitrate to N2 (Table 2 and Table S1). Considering the large human inputs of available nitrogen to the Earth's land surface, humans appear to have caused a surprisingly small change in the flux of N2 to the atmosphere from denitrification (17 TgN/yr). The efficiency of denitrification in streams appears to decrease as nitrate concentrations increase (94). Efforts to increase the rate of denitrification, while minimizing the byproduct flux of N2O, may be effective ways to reduce the environmental impact of fertilizer nitrogen (95, 96). Clearly even the modest flux of nitrogen to groundwater is problematic in regions depending on this resource for drinking water. More efficient use of fertilizer and efforts to increase vadose-zone denitrification may protect groundwater resources for future generations.
The sink for nitrogen in trees and soils appears modest (9 TgN/yr), but large enough to support a significant, simultaneous global sink for carbon in these ecosystem components. In contrast, on fertilized agricultural lands, the enhanced sink for carbon in soil organic matter is often substantially discounted by the CO2 emissions during fertilizer production (97). Inadvertent nitrogen deposition in natural ecosystems has important adverse consequences (16, 98), but it may alleviate progressive nitrogen limitation to forest growth under high CO2.
The largest net sink for anthropogenic nitrogen, atmospheric transport from the terrestrial to the marine realm, reflects the importance of airborne pathways and transformations of nitrogen compounds, much of which is derived from combustion sources. Many of the deposited compounds impact the productivity and health of coastal and continental shelf ecosystems. Better management of emissions from agriculture will help preserve the integrity of coastal ecosystems and their fisheries.
Overall, our understanding of the nitrogen cycle and the development of effective policies to reduce inadvertent losses of anthropogenic nitrogen to the environment is analogous to our understanding of the carbon cycle in the late 1960s. Humans are adding nitrogen to the Earth's surface; we do not know where it all goes, but we do know that increasing concentrations of nitrogen in unexpected places will cause significant environmental damage that we will all learn to regret.
Acknowledgments
This article derives from the author's presentation of the Michel T. Halbouty Lecture at the 2007 annual meeting of the Geological Society of America in Denver, Colorado. The author thanks many colleagues for stimulating discussions on this topic during the past few decades and Amy Burgin, Jim Galloway, Peter Groffman, and Peter Vitousek for helpful reviews of the manuscript.
Footnotes
- 1E-mail: schlesingerw{at}ecostudies.org
-
Author contributions: W.H.S. designed research, performed research, contributed new reagents/analytic tools, analyzed data, and wrote the paper.
-
The authors declare no conflict of interest.
-
This article contains supporting information online at www.pnas.org/cgi/content/full/0810193105/DCSupplemental.
-
Freely available online through the PNAS open access option.
- © 2008 by The National Academy of Sciences of the USA
References
- ↵
- ↵
- Galloway JN,
- et al.
- ↵
- ↵
- ↵
- Solomon S,
- et al.
- Denman KL,
- et al.
- ↵
- ↵
- Wollast R,
- MacKenzie FT,
- Chou L
- Isermann K
- ↵
- ↵
- ↵
- Smil V
- ↵
- ↵
- Schindler FV,
- Knighton RE
- ↵
- Fritschi FB,
- Roberts BA,
- Rains DW,
- Travis RL,
- Hutmacher RB
- ↵
- ↵
- Aneja VP,
- et al.
- ↵
- ↵
- Brooks M
- ↵
- Stevens CJ,
- Dise NB,
- Mountford JO,
- Gowing DJ
- ↵
- ↵
- ↵
- Scavia D,
- Bricker SB
- ↵
- Duce RA,
- et al.
- ↵
- Pregitzer KS,
- Burton AJ,
- Zak DR,
- Talhelm AF
- ↵
- Hyvonen R,
- et al.
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- Heilman PE,
- Dao TH,
- Cheng HH,
- Webster SR,
- Christensen L
- ↵
- Koopmans CJ,
- Tietema A,
- Boxman AW
- ↵
- ↵
- Proe MF,
- Dutch J,
- Miller HG,
- Sutherland J
- ↵
- Buchmann N,
- Gebauer G,
- Schulze ED
- ↵
- ↵
- ↵
- ↵
- ↵
- Zak DR,
- Pregitzer KS,
- Holmes WE,
- Burton AJ,
- Zogg GP
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- Goodale CL,
- Lajtha K,
- Nadelhoffer KJ,
- Boyer EW,
- Jaworski RA
- ↵
- Pacala SW,
- et al.
- ↵
- ↵
- ↵
- Howarth RW,
- et al.
- ↵
- Caraco NF,
- Cole JJ
- ↵
- Schaefer SC,
- Alber AM
- ↵
- Green PA,
- et al.
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- Spalding RF,
- Exner ME
- ↵
- Nolan BT,
- Hitt KJ,
- Ruddy BC
- ↵
- Rupert MG
- ↵
- ↵
- Alley WM,
- Healy RW,
- LaBaugh JW,
- Reilly TE
- ↵
- Berner EK,
- Berner RA
- ↵
- Schlesinger WH
- Galloway JN
- ↵
- Schlesinger WH
- ↵
- ↵
- Walvoord MA,
- et al.
- ↵
- Graham RC,
- Himas DR,
- Wood YA,
- Amrhein C
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- Houlton BZ,
- Sigman DM,
- Hedin LO
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- Tobias CR,
- Macko SA,
- Anderson IC,
- Canuel EA,
- Harvey JW
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- US Geological Survey
- ↵
- ↵
- Leonte C,
- Ibanhez JS,
- Rocha C
- ↵
- Matson PA,
- Naylor R,
- Ortiz-Monasterio I
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
Citation Manager Formats
Article Classifications
- Physical Sciences
- Geology
- Biological Sciences
- Environmental Sciences