NATUREiVol46418 April 2010
NEWS & VIE S
the angular size ofthe supergiant star (deter­
mined from interferometric measurements),
Kloppenborg et al. 1 were able to measure
the motion of the disk relative to the super­
giant and - knowing the orbital motion of
the supergiant from spectroscopy - they
determined its velocity relative to the binary's
centre of mass. This velocity implies that the
supergiant is about 60% as massive as its larger
disk companion. Most recently, Hoard et a1. 6
analysed the systems energy distribution from
the ultraviolet to the far-infrared part ofthe
electromagnetic spectrum and found that it
was best fitted by three components: a cool
(about 300°C) disk approximately four times
larger than the Earth-Sun distance; an unusual
lower-luminosity supergiant star of spectral
type F (surface temperature ofabout 7,500 °C)
about 135 times the radius ofthe Sun but only
two to three times its mass; and a hot BS-type
(about 15,000°C) 'main sequence' star about
six times the Sun's mass.
This unusual low-mass supergiant (typical
F-type supergiants are more luminous and
have masses oflO-20 solar masses) results
from the evolution of a 6-7-s01ar-mass pro­
genitor star through the asymptotic giant
branch (AGB) phase - in which much ofits
original mass is lost - to a post-AGB stage
supergiant. Post-AGB supergiants are essen­
tially bloated, dying 1-3-s01ar-mass stars
that are short-lived, and go on to eject more
material and end up as white dwarfs9 • In this
model, the observed large dusty accretion disk
is explained by the hotter, originally lower­
mass companion 'capturing' the material
ejected by the bloated supergiant as it evolved
through the AGB and post-AGB stages.
Although the new observations l ,6 tip
the scale in favour of the low-mass model,
as described above, it is still possible that the
supergiant star is a 'normal' and very young
object of 10-20 solar masses, and that its com­
panion is a massive proto-stellar accretion disk
object3,7. Because post-AGB stars typically have
lower luminosities than do normal massive
supergiants, one way in which to discrimi­
nate between the two possibilities would be to
determine the exact luminOSity ofthe super­
giant from a more accurate measurement of
its distance. In addition, detailed spectroscopic
analyses to search for the chemical anomalies
of post-AGB stars (due to convulsions and
'dredge-ups' of processed nuclear material)
could help to determine the nature of the
supergiant.
Kloppenborg and colleagues' study! has
demonstrated the vital role that eclipses con­
tinue to have in astronomy, and how well new
technologies can assist in solving old problems.
To help to study E Aurigae's eclipse, professional
astronomers have teamed up with amateurs
and organized an international observing
programme called Citizen SkylO. Continued
observations throughout the current eclipse
(the middle ofthe eclipse will occur in July­
August 2010) should more narrowly constrain
the properties ofthe system - for example, is
there an opening at the centre ofthe disk? Then
e Aurigae can be used as a laboratory for explor­
ing and testing current astrophysical concepts
and theories, including those concerning rapid
stages of stellar evolution, binary-star evolu­
tion, and the structure and dynamics oflarge
accretion disks, • 7. Carroll, S. M., Guinan, E., McCook, G, & Donahue, R.
Edward Guinan is in the Department of
Astronomy and Astrophysics, Villanova
University, Villanova, Pennsylvania 19085, USA.
e-mail: edward.guinan@villanova.edu
CLIMATE CHANGE
Grazing and nitrous oxide
Stephen J. Del Grosso
Most emissions of nitrous oxide from semi-arid, temperate grasslands
usually occur during the spring thaw. The effects that grazing has on plant
litter and snow cover dramatically reduce these seasonal emissions.
Nitrous oxide (N20) is a greenhouse gas that
also affects atmospheric chemistry and levels of
stratospheric ozone. Agriculture is the primary
anthropogenic source, with both cultivated and
grazed soils being major contributors. High
grazing intensity is thought to increase N 20
emissions because grazing increases the rate
of nitrogen cycling, and faster cycling gener­
ally leads to higher nitrogen losses from the
soil-plant-animal system.
However, on page 881 of this issue, Wolf
et a1. ! present evidence that grazing decreases
emissions from semi-arid grasslands that expe­
rience soil freeze-thaw cycles during the spring
season. When these types ofgrasslands are not
grazed, most ofthe annual emissions usually
occur during the spring, when subsurface soil
layers are still frozen and melting snow satu­
rates the surface soil layers, creating anaerobic
conditions that facilitate denitrification, one
ofthe primary biochemical pathways ofN20
emissions.
Grazing, however, reduces above-ground
standing biomass and plant height, so less snow
is captured and soil water content during the
spring thaw is diminished (Fig. 1, overleaf).
Less surface biomass and snow cover also
reduce insulation and lead to more extreme
diurnal fluctuations of soil temperatures.
Reduced water input from melting snow and
colder temperatures inhibit the soil micro­
bial activity responsible for N 2 0 emissions.
Consequently, current methods that do not
account for how grazing intensity affects soil
environmental conditions and microbial activ­
ity probably overestimate emissions from some
grasslands where snow melt and freeze-thaw
cycles are important processes.
Emission rates ofN20 are typically meas­
ured using 'bottom-up' methods that involve
placing airtight chambers over the soil surface
1. Kloppenborg, B. et 01. Nature 464, 870-872 (2010).
2. GoodricKe, J. Phil. Trans. R. Soc. 73, 474-482 (1783).
3. Guinan, E. F. & DeWar!, L. Astron. Soc. Pacif. Conf. Ser. 2 ,
121-142 (2002).
4. Ussauer, J. J., Walk, S. J., Griffith, C. A. & Backman, D, E,
Astrophys. J. 465, 371-384 (1996).
5. Huang, S,-S, Astrophys. J. 141, 976-984 (1965).
6. Hoard, D. w., Howell, S, B. & Stencel, R E. Preprint at
http://arxiv.org/abs/1003:3694(2010),
Astraphys.J. 367, 278-287 (1991).
8. www,chara,gsu.edu/CHARA
9, Webbink, R F. NASA Cant Publ. 2384,49-59
(1985).
10, www,CitizenSky.org
and measuring the change in gas concentr
tion over short time periods (15-60 minut
or so). Emission rates vary substantially, bo
spatially and temporally, and many studies ar
limited because oflow sampling frequenc
(less than one sample per week) or by samplin
only d~ring the growing season, when plant~
are active. '
Wolf et ai. 1 made their observations od
Mongolian grasslands subjected to differen~
stocking conditions, ranging from ungrazed t~
heavily grazed. They measured emission ratesl
year-round every three hours using automated\
chambers at two sites, and weekly using manual,
chambers at eight sites. During the growing sea-\
son, emissions increased with grazing intensity'\.
at most of the sites, consistent with previous I
studies and with models commonly used to I
estimate emissions. But most of the annual 'I'
emissions from the ungrazed sites occurred dur- ,
ing the non-growing season. By contrast, only \
a small proportion of annual emissions (less 1
than 10% on average) occurred during the non - \
growing season at the heavily grazed sites. On an
annual basis, emissions from the ungrazed sites
were more than double those from the heavily
grazed sites. This shows that previous studies
that considered only growing-season emissions
could lead to faulty conclusions regarding the
effect ofgrazing on total N 20 emissions.
The methodology ofthe Intergovernmental
Panel on Climate Change (IPCC) is commonly
used to estimate NzO emissions from agricul­
tural activities. For grasslands, emissions are
assumed to increase linearly with nitrogen
supplied from animal waste and decomposing
plant residues. The evidence presented by Wolf
et al. suggests that this methodology probably
overestimates emissions from some heavily
grazed semi-arid grasslands. On the basis of
climatological criteria used to identify regions
 JEWS & VIEWS NATUREIVol46418 April 2010
I
11 Contrasts in grazed and ungrazed grassland, January 2008. These scenes are from
¥ongc,iia, where Wolf et al. 1 made their measurements of nitrous oxide (N,O) emissions. Grazing
re~luaes v'eg!:tation and snow cover, leading to less soil water being supplied by melting snow.
grazing-induced changes inhibit microbial activity during the spring, when it turns out that
of the annual N,O emissions from ungrazed areas occur (right).
tv"here snow transport and spring freeze-thaw
f:ycles are likely to be important processes, IPCC
filethodology might overestimate grazing­
related emissions from about one-third ofglo­
al temperate grasslands (which in total cover
area ofabout 10 million square kilometres).
This possibility does not necessarily imply
that IPCC methodology overestimates agricul­
tural N,O emissions in general: at the global
scale, there is evidence3.4that emissions calcu­
lated using 'top-down' methods on the basis
ofchanges in N,O atmospheric concentration
and sink strength are consistent with emissions
estimated using IPCC methodology. But the
new workl does reinforce the assertion that this
methodology does not necessarily represent
the spatial and temporal patterns ofemissions,
and more complex models that simulate the
processes controlling emissions (for example,
plant nitrogen uptake and microbial activity)
usually agree more closely with plot -level obser­
vations4 • In particular, process-based models
have the potential to represent how grazing
intensity, snow cover and microbial dynamics
interact to control nitrogen cycling and N 20
emissions from these types ofgrasslands. But
. even process-based models are limited, because
I it is difficult, for example, to properly represent
, how wind velocity and topography affect snow
movement and accumulation.
Because N20 is a major greenhouse gas, Wolf
and colleagues' study could suggest a means
of mitigating greenhouse-gas emissions by
increasing grazing intensity. But that possibil­
I ity must be qualified because it is unclear what
portions oftemperate arid grasslands are cur­
rently grazed at levels that minimize NP emis­
sions. In addition to N p, the effects ofgrazing
intensity on other factors (such as plant pro­
ductivity, vegetation community structure, soil
erosion, levels ofsoil organic matter, and meth­
ane emissions) must be taken into account 5,6.
From a greenhouse-gas perspective, methane
produced by gut fermentation in ruminant
grazers is particularly important: methane
emissions from this source form a large portion
of total greenhouse-gas emissions associated
with livestock production.
Wolf and colleagues' results l are consist­
ent with reports showing that grazing had
little impact during the grov.ing season but
decreased spring and annual emissions from
a short-grass steppe in Colorado?, and that, in
some cropped systems, most ofthe annual NzO
emissions occur during the spring thaw, before
fertilizer applicationS. Clearly, there is a need
to measure NzO year-round in ecosystems in
which snow accumulation influences soil tem­
perature and hydrology. Increased microbial
activity in the ungrazed sites that accumulated
snow during the non-growing season is con­
sistent with the higher rates ofnitrogen cycling
observed during the winter in high -elevation
CANCER
Drug-tolerant insurgents
Paul Workman and Jon Travers
Some cancer cells that become tolerant to a drug remain resistant even
after its withdrawal t yet these cells eventually become sensitive to the drug
again. The underlying molecular mechanism is unusual.
The cell population ofa tumour is not uniform,
the differences arising partly as a result of
genetic instability. The prevailing model for
how cancer develops is that dominant cell
clones in the population survive through
Darwinian selection l - 3• But this donal selec­
tion model is based solely on genetic muta­
tions and, although useful, is overly simplistic.
It is not just the accumulation ofmutations in
cancer-causing genes that drives the lethal
properties of cancer4•5• Other factors that con­
tribute to cancer include the regulation ofcan­
cer genes by epigenetic means, the self-renewal
ofpotential cancer stem cells, and interactions
between tumour cells and their micro-environ­
mene,6. In addition to these intrinsic drivers of
cellular proliferation and survival, drug treat­
ments provide further selective pressure, lead­
ing to cells becoming drug resistant through
sites with deeper snow compared with low­
elevation sites with less snow in forests in the
northeastern United States9• By contrast, exper­
imental snow removal in these forests did not
affect nitrogen cycling, but did increase root
death rates and soil nitrate concentrationslO•
Overall, these results 1,7-10 show that generali­
zations about how changes in climate and land
management affect element cycling are often
confounded. Better projections will require
an improved understanding and modelling
ofthe processes that control biogeochemical
cycles. •
Stephen J. Del Grosso is at the US Department
of Agriculture, Agricultural Research Service,
Soil Plant Nutrient Research Unit, Fort Collins,
Colorado 80526, USA.
e-mail: delgro@nrel.colostate.edu
1. Wolf, B.etal. Nature 464, 881-884 (2010).
2. Eggleston, $" Buendia, L, Miwa, K., Ngara, T &
Tanabe, K. (eds) 2006IPCC Guidelines for National
Greenhouse Gas Inventories 10.53-11.54 (lnst. Global
Environmental Strategies, 2006).
3. Crulzen, P.1., Mosier, A. R., Smith, K. A. &Winiwarter, W
Atmos, Chem. Phys. 8,389-395 (2008).
4. Del Grosso, S. J., Wirth, T., Ogle, S. M, & Parton, W. 1.
Eos 89,529-530 (2008).
5. Derner,J D. & Hart, R. H. Rangeland fcol, Mgmt 60,
270-276 (2007).
6. liebig, M. A. Gross, J. R., Kronberg, S, L, Phillips, R. L. &
Hanson, J. D. L Environ. Qual. doi:l0.2134/jeq2009.0272
(2010).
7. Mosier, A R. et al. Global Biogeochem. Cyrlesn, 29-42 (1997).
8. Johnson, J. M, F., Archer, D. & Barbour, N. Soil Sci, Soc. Am. 1.
74,396-406 (2010).
9. Groffman, P. M. etal. Ecosystems 12, 927-943 (2009).
10. Groffman, P. M. etal. Biogeochemistry 56, 135-150
(2001),
both genetic and non-genetic mechanisms7,8.
Against this background ofthe existing mod­
els for cancer development and drug resistance,
in the latest issue of Cell Sharma et al. 9 report
a surprising mechanism for the emergence of
drug-resistant cancer cells. Their findings are
not just significant for our basic understanding
of cancer biology but are also ofconsiderable
importance for the treatment ofpatients.
Their story begins with the intriguing obser­
vation that, when a human lung-cancer cell line
was treated with the anticancer drug erlotinib
(Tarceva), most cells rapidly died, but a small
number persisted. This drug-tolerant popula­
tion' which the authors call drug-tolerant per­
sisters (DTPs), developed spontaneously from
single drug-sensitive clones, and the tolerance
was eventually lost after drug withdrawal.
Moreover, the authors showed that this form
Vol 46418 ApriI2010Idoi:10.1038/nature08931 na

LETTER
Grazing-induced reduction of natural nitrous oxide
release from continental steppe
Benjamin Wolfl, Xunhua Zheng 2, Nicolas BrOggemann l , Weiwei Chen 2 , Michael Dannenmann l , Xingguo Han 3,
Mark A. Sutton 4, Honghui Wu 3 , Zhisheng Yao 2 & Klaus Butterbach-Bahl l

Atmospheric concentrations of the greenhouse gas nitrous oxide
(N2 0) have increased significantly since pre-industrial times owing
to anthropogenic perturbation of the global nitrogen cycle1.1, with
animal production being one of the main contributors3 • Grasslands
cover about 20 per cent of the temperate land surface of the Earth
and are widely used as pasture. It has been suggested that high
animal stocking rates and the resulting elevated nitrogen input
increase NzO emissions....7 • Internationally agreed methods to
upscale the effect ofincreased livestock numbers on NzO emissions
are based directly on per capita nitrogen inputs8• However, mea­
surements of grassland NzO fluxes are often performed over short
time periods9, with low time resolution and mostly during the
growing season. In consequence, our understanding of the daily
and seasonal dynamics of grassland N2 0 fluxes remains limited.
Here we report year-round NzO flux measurements with high and
low temporal resolution at ten steppe grassland sites in Inner
Mongolia, China. We show that short-lived pulses ofNzO emission
during spring thaw dominate the annual NzO budget at our study
sites. The NzO emission pulses are highest in ungrazed steppe and
decrease with increasing stocking rate, suggesting that grazing
decreases rather than increases N 2 0 emissions. Our results show
that the stimulatory effect of higher stocking rates on nitrogen
cycling4,7 and, hence, on N2 0 emission is more than offset by the
effects of a parallel reduction in microbial biomass, inorganic
nitrogen production and wintertime water retention. By neglecting
these freeze-thaw interactions, existing approaches may have sys­
tematically overestimated N2 0 emissions over the last century for
semi-arid, cool temperate grasslands by up to 72 per cent.
In temperate ecosystems with long frost periods, distinct freeze­
thaw periods can occur. These periods can contribute significantly to
annual N 20 budgetslO~12 and need to be better understood in relatiop
to alterations in land use and agricultural practice. To address thesF
interactions, we deployed several N20 flux measurement systems i~
the steppe grassland of Inner Mongolia. Our measurement strate
was designed to address both sub-daily and seasonal dynamics, an
to consider the relationships between ~20 fluxes, grassland manage
ment and the underlying plant and soil processes.
1
The data set we report represents year-round (August 2007t
August 2008) NzO flux measurements at high and low temporajl
resolution at ten sites differing in stocking rate (Table 1). All sitef
are typical steppe grasslands belonging to the Inner MOngOj·'
Grassland Ecosystem Research Station, Chinese Ecosyste
Q
Research Network (43 33' N, 116° 42.3' E). At two of the sites,
fluxes were measured every three hours using an automatic systerri
combining triplicate automatic chambers with online analysis. Thes~
two sites compared grassland that had been ungrazed since 199~
(lJG99) with grassland that was grazed only during winter (WG)
At the remaining eight sites, manual replicated chambers werj
deployed, allowing samples to be taken weekly. Supporting measure
ments included soil temperature and water-filled pore space (WFPS).
gross rates of nitrogen mineralization and soil microbial-biomasS
nitrogen. I
Our automated NzO flux measurements show that NzO fluxes a~
both the ungrazed and the winter-grazed sites were low and dose to th~
detection limit during both the growing season and most ofthe winter.\.
However, at the end ofthe winter, during the spring thaw, a large pulse
ofN20 emissions, lasting for approximately eight weeks, was observed
at the ungrazed site. Contrary to expectations, there was no matching!
pulse of NzO emissions at the winter-grazed site (Fig. la, b). our\
sub-daily measurements also showed pronounced diurnal variations,

Table 1 Annual and seasonal cumulative fluxes of

Site Number of Grazing intensity'; stocking Duration of AE s.e, SE:;: s.e, Group mean of GE:;:s.e. Ratio of SE
chambers rate (sheep per hectare) grazing period (d) (kgNha- 1) (kgNha~l) SE:;: 5.e. (kg N ha~l) (kgNha~l) to AE (%)
UG99 3 Ungrazed; °° 0,22 0.D7 0.15 ± 0,04 0.17 O.Olt 0.06:: om
0,02 ± 0.01
68
UGI
UG2
L1
3
3
4
Ungrazed;
Ungrazed;
Light; 0.92
° 93
0.28 ± 0.05
0.17::: 0.03
0.20 0.D3
0.23 0,06
0.11 ± 0,03
0,15 ± 0.D3 0.10 0.01 H
0.04 0.01
om ± 0.01
81
66
77
L2 4 Light; 051 93 0,10 ::: 0.02 0.03 0.1 0,04 ± 0.02 35
Ml 4 Moderate; 1.45 93 0,15 ± 0.05 0.09 ± 0.03 0.06 ± O,Olt§ 0.04 0.05 60
M2 4 Moderate; 0,99 93 0,11 0.04 0.04 ± 0.02 0.05 ± 0.01 34
WG 3 Heavy; 2,05 166 0,01 :;: 0.03 -0.01 0.003 0.01 ::: 0,01§ 0.03 1. 0.01 °
HI 4 Heavy: 2.24 93 0.12 ± 0.01 0.02 ± om 0.06 ± 0,02 16,
H2 4 Heavy: 1,94 93 0.17 0.03 001 ± 0.001 0,11 ± 0.03 9.
N 3 for UG99, UGi. UG2, WG; N 4 for other sites. AE, annua: emission; GE, growing-season emiSSion; SE. spring-thaw emission. I
* Grazing intensity dasses derived from herbage allowance, Herbage allowance (HA; kiiograms of dry biomass per kilogram of life weight) is a measure of the amount of biomass available per mass,

of grazing animaL Classes are as follows. heavy grazing (HA < 25); mode'ate grazing (25 < HA < 7.5); grazing (HA > 7.5).

ttl Quantities marked with different symbols differ significantly (P< 0,05) with respect to gra2ing

for Meteorology and Climate Research, Karlsruhe Institute of Technology, Kreuzeckbahnstrasse 19, 82467 Garmisch-Partenkirche~. Germany.
Atmospheric Boundary Layer Physics and Atmospheric Chemistry. Institute of Atmospheric Physics. Chinese Academy of Sciences, 100029 Beijing. China. Laboratory of
Vegetation and Environmental Change, Institute of Botany, Chinese Academy of Sciences, 100093 Beijing, China. 4Centre for Ecology and Hydrology, Bush Estate, Penicuik EH26 OQB, i
UK,

881\
 ETTERS NATUREIVol46418 April 2010

~. r.§>'
,<:-oS ,,<J
1:
')'
"
E
01
.a
x
::>
""Z
0
z'"
E
E..
a'"
Cl

0
\L­
IE
::>
~
'"
0.
E
;2

~. o·
~ ,,<:)'Q "­
" Time (date in 2007-2008)

igure 1 I Dynamics of N20 fluxes, soil air concentrations and N= 3), and soil NzO concentrations are shown in parts per 109 (contours) as
nvironmental parameters. a-d, Steppe grassland ungra2ed since 1999 a function ofdepth in c and d. e, Soil environmental conditions, contrasting
(UG99; a, c) is contrasted with a winter-grazed site (WG; b, d). Measured temperature (solid lines) and WFPS (dashed lines) for UG99 (black) and
uxes of nitrogen from NzO (N2 0-N) are shown in a and b (error bars, s.e.; WG (red).

ofN2 0 fluxes at the ungrazed site during this period (Supplementary also found a significant positive correlation (P< 0.05) between the
Fig. 2). stocking rate and the contribution ofthe growing-season emissions to
Similar results highlighting the influence of grazing were also the annual N 2 0 budget. However, because the effect of the stocking
obtained at the sites where manual NzO flux measurements were rate in suppressing spring-thaw emissions was much more pro­
made. Pulse emissions of N 2 0 were most pronounced at sites that nounced, our results demonstrate that grazing decreased rather than
were either ungrazed or grazed at low intensity, but were hardly increased annual NzO fluxes.
detectable or not detectable at higher stocking rates (Fig. 2). Weekly measurements ofNzO in the gas profile showed that soil
Correlation analysis revealed that N 20 fluxes during spring thaw were N 2 0 concentrations increased only at the start of soil thawing.
significantly negatively correlated with stocking rate (1 0.57 (co­ Therefore, the observed peak NzO emission during spring thaw
efficient of determination), P< 0.05; Supplementary Information). was not due to an accumulation of N 2 0 in the profile during the
The spring-thaw N 2 0 pulses dominated the total annual N 2 0 emis­ preceding ...-inter. Measurements ofN2 0 production rates in strati­
sion and on average accounted for 72% of the annual emission for fied soil samples further support the notion that spring-thaw N 2 0
ungrazed sites and 8% of the annual emission for the heavily grazed fluxes were the result ofinstantaneous microbial production (Fig. Ic,
sites (Table 1). Other studies show that increasing stocking rates d), with the main production ofNzO occurring in the topmost layer
stimulate soil N 20 emissions during the growing season4-7. These (Supplementary Figs 3 and 4). Therefore, grazing must have changed
reports are not in contradiction to our findings, as in our study we the soil and environmental conditions that determine emissions of
NzO during the spring thaw. With increasing stocking rate, the
aboveground biomass and vegetation height decreased at our sites 13 •
Vegetation height is the determining factor in snow-holding capa­
0.30 -UG1
city, such that snow is more quickly eroded at grazed sites with sparse
'6.r:. --L1
or low vegetation than at sites with denser and taller vegetation l4 ,
0.25 --M1
z as also observed in our study (Supplementary Table 1). Moreover,
--H1
1.
z
0.20 we found that the grazed sites with lower vegetation/snow cover
have significantly higher freezing rates (Wilcoxon rank-sum test,
~ 0.15 P < 0.05; Supplementary Fig. 6) and lower ,~-inter soil temperatures
x
::> but increased soil temperature fluctuations on daily and annual
'+= 0.10
.~ scales. The decreased snowpack (analysis of variance, P< 0.05) at
1;;
:; 0.05 the grazed sites was associated with reduced soil moisture during
E snow melt (Fig. Ie). At site UG99, WFPS values were on average
8 0.00
61 % in spring (March to April) 2008, whereas at site WG, WFPS
-O.05+~-.---r-'---.--.--.---r--.---.--..--.-- was on average 37%. Accompanying measurements ofsoil microbial­
1 Aug. 1 Oct. 1 Dec. 1 Feb. 1 Apr. 1 Jun. 1 Aug. biomass nitrogen showed that at UG99 and WG, microbial popula­
Time (date in 2007-2008) tions decreased by 80% with the onset of winter. However, the
Figure 2 I Effect of stocking rate on cumulative NzO fluxes, as recorded recovery of the microbial biomass during the winter was faster at
using the manual-chamber approach. Error bars show uncertainty over the UG99, most probably because soil temperatures were not as low as
time span between two consecutive measurements. Uncertainty calculations at WG. In addition, gross nitrogen mineralization at the ungrazed site
were based on the spatial variability ofobserved fluxes (s.e.; N = 4 for LI, Ml was significantly higher than at the winter-grazed site (Supplemen­
and HI; N= 3 for UGl). Further site information is provided in Table L tary Fig. 1).
882
NATURE!VoI464!8 April 2010
The dominating role of soil moisture in controlling soil nitrogen
cycling and 1\20 losses is indicated by strong positive correlations
between soil water content and both microbial-biomass nitrogen
(? 0.90, P<O.OOl at WG; ?=0.81, P<O.OOI at UG99) and
1\20 emissions. The correlations between NzO fluxes and soil mois­
ture were weaker but were still significant (? = 0.11, P < 0.001 at
WG; ? = 0.32, P< 0.001 at UG99).
On the basis of our results, a general mechanism can be identified
that drives the interaction between grazing and freeze-thaw NzO
emissions. It is well established that tall vegetation accumulates snow"
and that, with increasing snow cover, the soil is effectively insulated
from low winter temperaturd s. The temperature effect is of import­
ance for supporting microbial grmrth during winter, because a critical
threshold seems to exist at approximately -10 °C (ref. 16). Above this
temperature, a vital microbial community can survive, and is further
activated by the onset ofthawing inspring'7. Our data, as well as earlier
studies'8,l9, demonstrate that freeze-thaw NzO emissions increase
with increasing soil moisture values and are highest at a WFPS value
of around 60%. Our explanation that snovvpack temperature insu­
lation and soil moisture effects are driving freeze-thaw NzO pulses is
also supported by a study at a prairie site in Colorado, USA20 • Relative
to a grazed site, the site with re-established grassland vegetation
showed larger winter NzO emissions, quantitatively consistent with
our own findings for Asian steppe (Supplementary InformationYo.
To investigate whether our finding of grazing-induced reductions
in N 20 emission could be meaningful on a global scale, we identified
grazed grasslands 21 that are climatically similar (data set CRU CL 2.0;
ref. 22) to our study sites and the prairie site in Colorad0 20 • For this,
we defined semi-arid, cool temperate grasslands as follows: at least
90% of the days during three consecutive winter months have frost
(for sensitivity analysis we also used values of 92.5% and 95%);
winter precipitation is >5 mm; annual precipitation is <600 mm.
We further excluded North American tallgrass prairie as it may cap­
ture snow even if grazed (Supplementary Figs 7 and 8).
For the selected grasslands (29-35% of grasslands in temperate
zones), we calculated historic and recent livestock-related N 2 0 emis­
sions using internationally agreed methodologf (Supplementary
Information) and tested two possible situations: in the first (Sl),
we considered the effects of livestock numbers and management
practices, without application of synthetic nitrogen fertilizer, this
being representative of many developing countries; in the second
(S2), we considered the additional effect offertilizer application rate.
In S2 we assumed that application ofsynthetic fertilizer to pastures in
temperate grasslands started in 1951 and increased linearly to a value
corresponding to 23kgNha- 1 yr- 1 (ref. 23). We did not consider
run-off of nitrate and downstream production ofN 2 0. We have not
observed this process at our sites, but cannot exclude its occurrence
elsewhere.
The main driving factor in our calculations was the numbers of
grazers. We calculated livestock-related ]\;"20 emissions for 1890­
2000 in two ways per situation (Fig. 3): on the basis of the number
of grazing animals (cattle, buffalo, goats and sheep)24--27 according to
the approach of the IPCC8 ; and considering the decreasing effect of
increasing intensification of livestock farming (expressed by animal
number) on background spring-time N 20 emissions, as demonstrated
in our measurements (Table 1; see Supplementary Information for the
calculation methodology). These estimates suggest that, according to
the IPCC methodology., the N 20 emission from semi-arid, cool
temperate pasture systems could be overestimated by 72% (Sl) Of, if
fertilization is considered, 36% (S2).
The measurements presented here reveal a mechanism that may be
relevant to grasslands in semi-arid, cool temperate regions in generaL
Our results are therefore likely to be pertinent in the context of
the current debate on the global magnitude ofN 2 0 emissions from
agriculture 3 ,28. The anthropogenic modification of so-called 'back­
ground fluxes', which we report here, demonstrates the complexity
and high degree of uncertainty in constraining the global 1\20
0.14
~ 0.12 ,
c
I
0
'0;
0.10
'"
E
1 x 108
(])
0.08 ~
<;:
ON
z 0.06 ~
OJ
OJ
c
c 0.04
0,5x108~
«
0.02
1880 1900 1920 1940 1960 1980 2000
Year
Figure 3 I Annual N 2 0 emissions and livestock numbers between 1890 and
2000 for situation 51. N 2 0 emissions due to increases in animal numbtrs
determined using the default methodology of the Intergovernmental Pa,.el
on Climate ChangeS (IPCC; black); livestock numbers (cattle, buffalo, sh<jCp
and goats; blue); and N 2 0 emissions corrected for the decrease in spring­
thaw fluxes as calculated here (red). The traces and the grey shaded area$
represent mean :t s.d. of calculations for different frost-day climate criteria
(90, 92.5 and 95% frost days in three consecutive winter months). The
shaded areas thus indicate differences in the results due to changes in th~
extent of the projection area. The decrease in animal numbers since 19801 is
due to more efficient livestock farming in developed countries.
budget, highlighting the need to develop more sophisticated upscal­
ing approaches. As our measurements in the steppe of Innfr
Mongolia show, human intervention-here grazing of livestockr'
has the potential to reduce natural background K10 fluxes signi ­
candy. An understanding of these controlling mechanisms also
provides pointers to identifying improved N 20 mitigation strategids.
In addition to the effects of grazing, it is likely that cutting arjd
haymaking at the end of the growing season have the potential to
decrease spring-thaw N10 emissions owing to the above-mentiondd
effect of vegetation height on snow accumulation. Although Hie
possible trade-offs with water balance, snow drift and grass produ<t­
tivity would need to be quantified for different landscape type~,
such simple management practices could have a significant role in
reducing N 2 0 emissions from temperate grasslands on a global scal¢.
METHODS SUMMARY
In the automatic-chamber system, NzO analysis was conducted using an onli:de
gas chromatograph with electron capture detection (SRI 8610C, Tex4s
Instruments), calibrated against NzO standards with concentrations of 400 pa~s
per 109 by volume (Air Liquide). In the manual-chamber system, air sampl~s
were collected in 100-ml syringes and analysed the same day by gas chromat9­
graphy (6820D, Agilent Technologies). We calculated fluxes from the increase ij1
N 20 concentration in the chamber and corrected them for atmospheric pressure
and chamber air temperature. Each site was equipped with at least three cham~
bers of at least 40 em X 40 em base area. The flux detection limit was <2 J.lg N20~
N m -2 h - ,. We measured soil N 2 0 concentrations using gas-permeable tubes aJt:
different depths.
The 15N pool dilution technique'· was applied to measure gross rates of
nitrogen mineralization (ammonification) at three replicated plots for each or
the sites UG99 and WG at fortnightly to monthly intervals. Microbial-biomas~
nitrogen was determined by use of the chloroform-fumigation extractio~
method". We determined the microbial biomass and related parameters withi1
the top 10 em of the soil. Soil temperature was determined at depths of 1, S, HI
and 20 em, and soil moisture was determined in the top 5 em, '
For statistical analysis ofN 20 fluxes from the ten sites during spring thaw, w4
aggregated UG99 and WG data to the same measurement frequency as the site4
with low temporal measurement resolution (n 7).In the case ofnon-normalit:)f
(Shapiro-Wilk test) and different variances (Bartlett test) of the N 2 0 flux datal
we used multiple pairwise Wilcoxon tests to test for significant differences i~
spring-thaw fluxes between grazing intensities. We corrected Pvalues for mlll~
tiple testing. The Wilcoxon test was further applied to test for significant differ,
ences in microbial-biomass and gross nitrogen mineralization between UG9~
andWG, !
883,
L TTERS
R ceived 28 April 2009; accepted 15 February 2010.
1. Solomon, S, et ai, (eds) Climate Change 2007: The Physical Science Basis 143-145
(Cambridge Univ, Press. 2007).
2. Mosier, A et a/, Closing the global N20 budget: nitrous oxide emissions through
the agricultural nitrogen cycle - OECD/IPCC/IEA phase II development of IPCC
guidelines for national greenhouse gas inventory methodology. Nutr. Cye/,
Agroecosyst. 52, 225-248 (1998).
3. Davidson. E. A. The contribution of manure and fertilizer nitrogen to atmospheric
nitrous oxide since 1860. Nature Geosci. 2, 659-662 (2009).
4 Ma. X. Z, et 01. Short-term effects of sheep excrement on carbon dioxide, nitrous
oxide and methane fluxes in typical grassland of Inner Mongolia. N. Zeal. 1. Agric.
Res. 49, 285-297 (2006).
5 Oenema. 0 .• Velthol, G. L., Yamulki, S. & Jarvis, S. e. Nitrous oxide emissions from
grazed grassland. Soil Use Manage. 13, 288-295 (1997).
6 Saggar, S., Hedley, e. B., Giltrap, D. L. & lambie, S. M. Measured and modelled
j estimates of nitrous oxide emission ahd methane consumption from a sheep-
grazed pasture. Agdc. Ecosyst. Environ, 122, 357 - 365 (2007).
Yamulki, S., Jarvis, S. e. & Owen, P. Nitrous oxide emissions from excreta applied
in a simulated grazing pattern. Soil Bioi. Biochem. 30, 491-500 (1998).
Eggleston, S. Buendia, L, Miwa. K., Ngara, T. & Tanabe, K. (eds) 2006IPCC
Guidelines for National Greenhouse Gas Inventories 10.53-11.54 (institute for Global
Environmental Strategies. 2006).
Matzner, E. & Borken, W. Do freeze-thaw events enhance C and N losses from
soils of different ecosystems? A review, Eur. 1. Soil Sci. 59,274-284 (2008),
t
1.'
.
Groffman, P. M., Hardy, J. P., Driscoll, C T. & Fahey, T. J. Snow depth, soil freezing,
and fluxes of carbon dioxide, nitrous oxide and methane in a northern hardwood
forest. Glob. Change Bioi. 12, 1748-1760 (2006).
1[" Papen, H. & 8utterbach-Bahl, K. A 3-year continuous record of nitrogen trace gas
fluxes from untreated and limed soil of a N-saturated spruce and beech forest
ecosystem in Germany - 1. N2 0 emissions. 1. Geophys, Res. 104, 18487-18503
(1999).
1 . Rover, M., Heinemeyer, 0, & Kaiser, E. A. Microbial induced nitrous oxide
emissions from an arable soil during winter. Soil Bioi, Biochem. 30, 1859-1865
~:?t~~nn,
1r', 01.
C et Effects of grazing and topography on dust flux and
deposition in the Xilingele grassland, Inner Mongolia. J. Arid Environ. 72, 792-807
(2008).
1~. Essery, R. & Pomeroy, J. Vegetation and topographic control of wind-blown snow
distributions in distributed and aggregated simulations for an Arctic tundra basin.
J. Hydrom. 5,735-744 (2004).
Brooks, P. D., Williams, M. W. & Schmidt, S, K, Inorganic nitrogen and microbial
biomass dynamics before and during spring snowmelt. Biogeochemistry 43, 1-15
(1998).
Rivkina, E, M., Friedmann, [I., McKay, e. p, & Gilichinsky, D. A Metabolic activity
of permafrost bacteria below the freezing point. Appl, Environ. Microbial. 66,
3230-3233 (2000).
Author Information Reprints and permissions information is available at
Sharma, S. et al. Influence of freeze-thaw stress on the structure and function of
www.nature.com/reprints. The authors declare no competing financial interests.
microbial communities and denitrifying populations in soil. Appl, Environ.
Microbial. 72,2148-2154 (2006).
84
NATUREIVol46418 April 2010
18. Koponen, H. T & Martikainen, P. J. Soil water content and freezing temperature
affect freeze-thaw related N 20 production in organic soil. Nutr. Cyel, Agroecosyst.
69, 213- 219 (2004),
19. Teepe, R" Vor, A, Beese, F. & ludWig, B. Emissions of N2 0 from soils during cycles
of freezing and thawing and the effects of soil water, texture and duration of
freezing. Eur. J. Soil Sci. 55, 357-365 (2004).
20. Mosier, A R. et ai, CH 4 and N 2 0 fluxes in the Colorado shortgrass steppe. 2. long­
term impact of land use change. Glob. Biogeochem. Cycles 11, 29-42 (1997).
21. Ramankutty. N., Evan, A. T, Manfreda, e. & Foley. J. A, Farming the planet: 1.
Geographic distribution of global agricultural lands in the year 2000. Glob.
Biogeochem. Cycles 22, GB1003 (2008).
22. New, M., lister, D., Hulme, M. & Makin, I. A high-resolution data set of surface
climate over global land areas. Clim. Res. 21,1-25 (2002).
23. FertiStat. Fertilizer use by crop statistics. Food Agric. Or9, UN (http://
www.fao.org/ag/agl/fertistat/index_en.htm) (2007).
24. FAOST AT. Food Agric. Org. UN (http://faostat.fao.org/site/573/
default.aspx#ancor) (2008).
25. Mitchell, B. R. International Historical Statistics: Europe 1750-1988 3rd edn,
325-370 (Stockton, 1992).
26. Mitchell, B. R. International Historical Statistics: The Americas 1750-1988 2nd edn,
Ch. C (Stockton, 1993).
27. Mitchell, B. R. International Historical Statistics: Africa, Asia and Oceania 1750-1988
2nd rev. edn, Ch. C (Stockton. 1995).
28, Crutzen, P. J., Mosier. A R., Smith, K. A. & Winiwarter, W. N 2 0 release from agro­
biofuel production negates global warming reduction by replacing fossil fuels.
Atmos. Chem. Phys. 8,389-395 (2008).
29. Dannenmann, M" Gasche, R., Ledebuhr, A & Papen, H. Effects of forest
management on soil N cycling in beech forests stocking on calcareous soils. Plant
Soil 287, 279-300 (2006).
Supplementary Information is linked to the online version of the paper at
www.nature.com/nature.
Acknowledgements This work has been supported by the German Research
Foundation (DFG research group 536, 'Matter fluxes in grasslands of Inner
Mongolia as influenced by stocking rate') and the National Natural Science
Foundation of China (grant number 40805061), with co-funding from the
NitroEurope Integrated Project of the European Commission. We thank
K. K. Goldewijk for providing livestock data for the years before 1961, and Z. Yu,
K. MOlier, L lin, P. Schoenbach, G. Willibald, R. Kiese. e. Werner and C. Liu for
support with field measurements.
Author Contributions K.B.-B., N.B., XL and M.D. designed the experiment. B.W.,
W.e. and Z.Y. carried out the flux measurements. H.W. conducted the
microbiological measurements. B.W., W.e., H.W. and M.D. performed data
analysis. B.W. carried out the upscaling for temperate grasslands. MAS., K.B.-B.,
N,B., M.D. and 8,W. drafted the manuscript.
Correspondence and requests for materials should be addressed to K.B.-B.
Cklaus.butterbach-bahl@kit.edu).