 

gg £3

 

kg
«if
. a. u . . ‘ V . , s, ,
a»; «1% $1”: 3
I. u .u‘
.r

3 v. ‘
.

3.3: . . E

5“ .
mm. $4 ﬁ.
w.“ m3 WWW.

:4 . . .. a". "may”
eﬁwhﬁmmwnz . :50 a 3.. WNW .#
.Mw 33$

:1 .23.! ..
.. an 2.2539
in ~ a»

9.!
t;
.1:
~
I .

Ev

a:

..
43%“.
.s

_ .wu
.u.
. “NI-3‘“

:5 .

in,“ 534-! 1 .135-
.75 2.

Wsnﬁn .5 ‘ T ,
u‘ .. mm“?
1...: ,ﬂhymu; .

 

 

 

$
4"
u ‘ . , A »
,. Jar... . , .A d».
,. av

‘. ‘ é. . ”Emu
‘ . 53v .K d...

. V ‘2 . . I u
.Pé....£.:ﬂ. . n

.5. I
‘r

i a
TB . :
pi”

i

I
535%?

. 43...”:
s 1 41:35

 

 

2cm

This is to certify that the

dissertation entitled

THE RELATIONSHIP BETWEEN DENITRIFICATION
AND NITROUS OXIDE FLUX FROM SOIL

presented by
Timothy Todd Bergsma

has been accepted towards fulﬁllment
of the requirements for

 

 

Ph. D. degree in Ecology

(DMM

M ajorHesTsor

Date 9— mm:

 

MS U is an Afﬁrmative Action/Equal Opportunity Institution 0- 12771

 

LIIRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

”a 9.; 553
SEP 0 4 2007

11;

”9132832

 

 

 

 

 

 

 

 

 

 

 

 

 

11/00 c/CIRC/Dateotnpsfm,“

 

 

 

THE RELATIONSHIP BETWEEN DENITRIFICATION AND NITROUS OXIDE
FLUX FROM SOIL

By

Timothy Todd Bergsma

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Crop and Soil Sciences

2000

ABSTRACT

THE RELATIONSHIP BETWEEN DENITRIFICATION AND NITROUS OXIDE
FLUX FROM SOIL

By

Timothy Todd Bergsma

Denitriﬁcation in soil is the major source of atmospheric nitrous oxide
(N20), a potent contributor to global warming and a regulator of stratospheric
ozone. Flux of N20 to the atmosphere is poorly understood. There is a serious
imbalance in the global N20 budget (missing sources), and N20 ﬂux at the ﬁeld
scale is difﬁcult to predict, even when the rate of denitriﬂcation has been
adequately characterized. In this dissertation, I review the literature pertaining
to the relationship between denitriﬁcation and N20 ﬂux, emphasizing a major
source of uncertainty: relative production of N20 and N2 (dinitrogen) during
denitriﬁcation. I explore the ecological factors that inﬂuence the nitrous oxide
mole fraction ( N201[ N20 + N2] ), an aspect of denitriﬁcation that expresses
relative production of N20 and N2. Second, I develop theory that facilitates the
evaluation of N20 and N2 ﬂux using 15N-labeled compounds and mass
spectrometry. My heuristic model of labeled N-gas flux from soil simpliﬁes the
process of drawing inferences from isotope data. Third, I report new
procedures for measuring ﬂuxes of labeled N20 and N2 for the same incubation

using mass spectrometry. My procedures could lead to improved estimates of

the nitrous oxide mole fraction. l illustrate these procedures with ﬁeld data.
Finally, I describe a laboratory experiment in which I combine traditional and
isotope methods to test for effects of moisture history (antecedent soil
moisture) and ecosystem management history on nitrous oxide mole fraction,
while controlling soil type and moisture. I ﬁnd that response of mole fraction to
differences in short-term (48 h) soil moisture history is different for soils from
ecosystems with different management histories. A cropped soil had a high
(~0.9) mole fraction after rapid transition from air-dry to 85% water-ﬁlled pore
space but a low (~0.3) mole fraction when 80% of added moisture was applied
48 h in advance of the incubation. However, soil from a successional system
generated a nitrous oxide mole fraction of about 0.3 regardless of short term
moisture history. Progress in understanding the relationship between

denitriﬁcation and nitrous oxide ﬂux from soil seems still to be methods-limited.

For Qiaobing. W0 ai ni!

ACKNOWLEDGMENTS

I thank God for creating a world that is both interesting and (often!)
intelligible. I thank Jesus for rescuing me from sin and for inspiring me, by his
Spirit, to pursue a life of service and gratitude. I thank my wife Qiaobing
(Crystal) who never complained when I came home from the lab later than I
said I would (which was always). I thank her also for giving me a marriage and
a family, and for picking up more than her share of household responsibility so
that I could ﬁnish the degree program represented by this dissertation. I thank
David for helping me with ﬁeld work (we have the pictures!) and I thank Maarten
for a thousand smiles. Special thanks are due to Yiyan Liang, my mother-in-
law, who cared for Maarten while I was finishing my lab work and writing. I
thank my parents for their encouragement and support.

Deepest thanks are extended to my graduate committee. My advisor,
Phil Robertson, artfully guided my degree program. He never pushed me for
results, which was just the right strategy to get me to push myself. I am deeply
grateful for all he has taught me about planning research and especially about
communicating it in writing. Nathaniel Ostrom spent uncounted hours teaching
me mass spectrometry and non-lethal use of a vacuum line. More importantly,
Nathaniel taught me the value of goal-directed problem solving by modeling a
deft economy of lab effort. Even without his training, this dissertation would
have been impossible apart from Nathaniel’s spectrometry skills. Thanks also

are due to honorary committee member Peggy Ostrom, whose distance from

the project and research savvy resulted in many critical suggestions that cut
through my technical or conceptual waywardness. Steve Hamilton was an
inspirational role model as well as effective critic of my experimental designs. I
thank him also for access to critical lab equipment and services. Eldor Paul
offered the seasoned perspective of an entire research community. Some of
the fundamental concepts of my research project grew out of his coursework
assignments. Appreciation is due also to the constellation of KBS faculty for
creating a nurturing environment for research.

I thank Per Ambus for introducing me to ﬁeld sampling and to data
management. Kevin Kosola conspired with me to contrive all manner of
beastly ﬁeld equipment; I admire his inventiveness and thank him for
encouragement. Heather Reynolds listened patiently to my ideas about spatial
variability of soil resources, and provided excellent guidance. Claire McSwiney
laughed and cried with me over the torments of understanding anything about
nitrogenous trace gases - a kindred spirit. Jane Boles was a sister in more
than a strictly academic sense; I thank her for her friendship and admire her
ability to think outside her ﬁeld. Michel Cavigelli was something of a hero, as
well as mentor and conﬁdant. Martha Tomacek was a pillar of emotional
support. Stuart Grandy and Pongthep Suwanwaree shared ideas and lab
space. I thank all the K88 graduate students - and not a few from Zoology,
Botany, and Crop & Soils - for friendship and collegiality. In particular, I thank

Michelle Pruyn for introducing me to Qiaobing.

Time and space fail me if I try to mention all the support staff that
contributed substantially. I thank the staffs - past and present - of the
Robertson, Hamilton, and Klug labs. Special thanks to Olivia Daemon, Kay
Baergen, Sandy Halstead, John Furguson, Christine Easley, Sandy Marsh, and
Madhav Machavaram. My ﬁeld assistant Jennifer Siekierski patiently executed
many bizarre - and sometimes futile - ﬁeld treatments for me, with the
courtesy of not voicing the obvious. I thank Barb Fox for handling innumerable
administrative details. Special thanks to Nina Consolatti, who took a genuine
interest in my equipment needs, and shared with me membership in the secret
circle of those who really know how the weather station operates. John Gorentz
cheerfully solved myriad computing problems, and shared my interest in the
process as well as the solution.

Randy De Jong, Nate Dom, and Greg Houseman were faithful friends;
they listened patiently to my sad stories when experiments were ﬂoundering
and gave much needed encouragement. I am inspired by their dedication and
enthusiasm. My REU student Mike Case had as much inﬂuence on me as I
had on him, I’m sure. I thank him for his energy, creativity, and persistence,
and for teaching me more about nitriﬁcation than I ever thought I wanted to
know. I thank Gabriel Vogeli for extensive help and advice on career
development.

Support from the following sources is gratefully acknowledged: the

USDA-NRI Program, the NSF RTG and LTER Programs, the Michigan

vii

Agricultural Experiment Station, and the C. S. Mott Fellowship of Sustainable

Agriculture.

viii

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................ xiii
LIST OF FIGURES .................................................................................................... xvii
INTRODUCTION .................................. 1
Chapter 1

FACTORS INFLUENCING THE RELATIONSHIP BETWEEN

DENITRIFICATION AND NITROUS OXIDE FLUX FROM SOIL. ........................... 3
Background ............................................................................................................... 3
Challenges for research ........................................................................................ 4
Factors influencing N20 mole fraction during denitriﬁcation ........................... 6

Theory .................................................................................................................... 6
Temperature ......................................................................................................... 9
Available carbon ................................................................................................... 9
pH ......................................................................................................................... 1 0
Nitrate and nitrite ................................................................................................ 11
Soil Water and Oxygen ...................................................................................... 12
Enzyme status .................................................................................................... 13
Time ..................................................................................................................... 13
Chapter 2

A HEURISTIC MODEL FOR THE CALCULATION OF DINITROGEN AND
NITROUS OXIDE FLUX FROM NlTROGEN-15-LABELED SOIL ....................... 20

Abstract .................................................................................................................... 21

Measuring ﬂux ........................................................................................................ 21

Graphical representation ..................................................................................... 21

Equaﬁons ................................................................................................................ 22

Expectation of error due to multiple pools ........................................................ 23

Discussion .............................................................................................................. 25

Notes ........................................................................................................................ 27
Chapter 3

A NOVEL METHOD FOR DIRECT DETERMINATION OF NITROUS OXIDE

AND DINITROGEN FLUX FROM 15N-LABELED SOIL ......................................... 29
Summary ................................................................................................................. 29
Introduction ............................................................................................................ 30
Materials and Methods ......................................................................................... 32

Overview .............................................................................................................. 32
Vessels and Sampling ..................................................................................... 33
Analysis ............................................................................................................... 34
Data Processing ................................................................................................ 36
Veriﬁcation ........................................................................................................... 37
Laboratory Denitriﬁcation ................................................................................. 39
Field Demonstration ......................................................................................... 41
Results .................................................................................................................... 42
Veriﬁcation ........................................................................................................... 42
Laboratory Denitriﬁcation ................................................................................. 43
Field Demonstration ......................................................................................... 44

Precision and Detection Limits ....................................................................... 45

Discussion .............................................................................................................. 47
Notes ........................................................................................................................ 52
Chapter 4

NITROUS OXIDE MOLE FRACTION DURING DENITRIFICATION IN SOIL:
RESPONSE TO RECENT MOISTURE HISTORY VARIES AMONG

ECOSYSTEMS ............................................................................................................ 72
Summary ................................................................................................................. 72
Introduction ............. . ............................................................................................... 73
Materials and Methods ......................................................................................... 74

Soil collection and processing ........................................................................ 74
Experiment and treatments ............................................................................. 75
Sampling and analysis ..................................................................................... 76
Results .................................................................................................................... 78
Experimental design ......................................................................................... 78
N20 mole fraction ............................................................................................... 79
Isotopic data ....................................................................................................... 80
Discussion .............................................................................................................. 82
Experimental Design ........................................................................................ 82
N20 mole fraction ............................................................................................... 84
Isotope data ........................................................................................................ 90
Conclusions ........................................................................................................... 94
OVERALL CONCLUSION ....................................................................................... 109

APPENDIX ................................................................................................................. 1 1 1

Literature Cited ......................................................................................................... 116

xii

LIST OF TABLES

Chapter 1
Table 1.1. Summary of factors potentially influencing N20 mole fraction.
The generally-accepted direction of the correlation is indicated.
WFPS = water-filled pore space. .................................................................... 17
Chapter 3
Table 3.1. Test for stability of measured ratios with varying sample size.
Variations were simulated by adjusting the strength of the major
beam for analysis of N20 laboratory standard gas. 2/1 and SH
refer to mlz ratios 45/44 and 46/44, respectively. SE is standard
error, n is number of samples. ....................................................................... 55
Table 3.2. Test for a memory effect during analysis. Highly enriched
N20 was prepared by mixing various quantities of labeled and
unlabeled N20. Analysis of laboratory standard N20 immediately
followed each analysis of enriched gas. Since the laboratory
standard is the same as the reference gas, a value of O %o is
expected if there is no memory effect. ........................................................... 56
Table 3.3. Summary of analyses of high-enrichment laboratory mixtures
of N20. 44:45:46 represents the mixing ratio of natural abundance
N20, “N20, and 46N20. Predictions 15a and 45N20 adjust for gas
purity. ‘53 is the atom fraction of 15N in the sample, consistent with

the notation of Arah et al. (1992). 45N20 is the molecular fraction of

xiii

mass 45 in the sample. Measured 15a and “N20 are means for

all samples where number of samples n is more than 1. ........................ 57
Table 3.4. Test for effects of the LiOH water trap, water in LiOH, and time

on measured enrichment of a 1:12 mixture of natural abundance

N20, “N20, and 46N20. 'Hydrous' refers to hydrous LiOH used as

a chemical trap, 'dried' refers to hydrous LiOH dried on a vacuum

line overnight. “N20 and 4‘5N2O represent the mean calculated

molecular fraction for N20 of masses 45 and 46. SE is standard

error. ..................................................................................................................... 58
Table 3.5. Test for an effect of collecting sample gas quickly using pre-

evacuated vessels. Air samples were collected by diffusion into

open vessels or by rapid ﬁlling of pre-evacuated vessels when

stopcocks were opened. Samples were analyzed for N20 using

the unmodiﬁed head ampliﬁer. 2/1 and 3/1 are mlz ratios of 45l44

and 46/44, respectively. Means :I: standard errors are reported; n

is 3. P > m is the signiﬁcance level for Student's t test of

differences in means. Least signiﬁcant number (LSN) is the

smallest number of samples needed to demonstrate significant

differences in means at a conﬁdence level of a = 0.05 .............................. 59
Table 3.6. 15N enrichment of soil N03' pools for laboratory denitriﬁcation

experiments, calculated and measured . Calculations are based

on the enrichments and mixing ratios of natural abundance KNOa'

and highly enriched KNOa' (99.93 atom %). Measurements use

the equations of Bergsma et al. (1999). 'n' is number of samples.
Means :I: standard errors are reported for separate experiments
measuring N2 and N20. P>|t| is the signiﬁcance level for Student's
ttest. ..................................................................................................................... 60
Table 3.7. Comparison of spectrometer precision by study. Precision is
reported for analysis N2 or N20. 2/1 refers to mlz 29/28 or 45l44.
3/1 refers to 30/28 or 46144, except as noted. STD is standard
deviation and n is number of samples .......................................................... 61
Chapter 4
The isotope data show that the mineral N pool undergoing denitriﬁcation
was isotopically uniform in most cases. There is no conclusive
explanation for the strong differences between estimates of N2
production by acetylene inhibition and by 15N isotope dilution.
Perhaps, under the experimental conditions described, there
existed an alternative substrate for production of N2, but not for
N20; the 15N method may then have reported only production of
labeled N2, while acetylene inhibition would have reﬂected gross
production of N2. Factors likely contributing to the difference
between methods are the complexity of the soil environment and
the dynamic nature of N transformations during rapid re-wetting of
soil. ....................................................................................................................... 95
Table 4.1. Nitrous oxide mole fraction ( N20 l [ N20 + N2] ) analyzed by

ecosystem and recent soil moisture history. Estimates are mean

:I: standard error, in pg N - g dry soil '1. Effects are P values (Prob.

> F). “Ecosystem by history” is the interaction term. .................................. 96
Table 4.2. Comparison, by ecosystem and replicate, of predicted soil

NO; enrichment (atom fraction 15N) with apparent enrichment of

the soil pool undergoing denitrification. “Predicted” is calculated

by mass balance from the extractable NO; levels in stock soil and

the known addition of KNO3. Apparent enrichments are from N2

isotope data (“by Ng”), from N20 isotope data (“by N20”), or

“average” of N2 and N20 estimates. ............................................................... 98
Table A1. Tests of signiﬁcance (of differences among means) for Figure

4.2: probabilities of ﬁnding the observed differences in means by

chance alone. Each row represents a single model of

“Response”, incorporating the effects for which there are column

entries. In most cases, models returning P > 0.05 for an effect

were re-run without that effect ("-"). Dropping the effect (“-") is a

notational convenience only; statistically it is no different from

grouping the responses across levels (“grouped"). Levels of

block are [4, 5, 6]. Levels of ecosystem are [cropped

successional]. Levels of history are [pre-wet, control]. “NA"

means the effect (column) is not meaningful for testing the

response (row) ................................................................................................. 1 12

LIST OF FIGURES

Chapter 1
Figure 1.1. Conceptual diagram: factors influencing ﬂux of N20
from soils. WFPS = water-ﬁlled pore space. C = organic
carbon availability. P = phosphorus availability. ............................. 18
Chapter 2
Fig. 1. Illustration of the equivalence of ternary and Cartesian
plots. (a): One thousand simulated N2 samples, with each
molecular fraction independently randomized, plotted as a
ternary graph. (b): These same one thousand samples
plotted as 29x vs. 15a. .............................................................................. 23
Fig. 2. Visualization of the heuristic model (Eq. [1], [2] and [3]). A
represents the atmosphere and Mrepresents the sample:
positions of both are determined analytically. The ray
drawn from A through M intersects the equilibrium curve,
implicating a pool P of gas derived from the soil, having an
equilibrium distribution of masses. ................................................... 24
Fig. 3. Visualization of underestimation due to multiple soil N
pools of differing enrichment. If P is in fact a mixture of
equilibrium gases (resulting from multiple pools) then it
Pcalculated

necessarily falls below the equilibrium curve. Thus

is necessarily an overestimate of Padua', when 15a,, > 15a...

Since the segment AP‘i‘k‘miW’“l is longer than AP‘W“, ﬂux is
necessarily underestimated (since ﬂux cc AM 1 AP) ......................... 25

Fig. 4. Visualization of a method to circumvent the problem of pool
non-uniformity. After an initial incubation in which A, and M
are sampled, the chamber is ﬂushed, closed, and spiked
with ”N2. A2 is displaced relative to A1. Measurement of N12
gives the trajectory of a second ray that converges on the
ﬁrst at Pam’a'. The principal assumption is constant isotopic
character of the evolving gas. (Scale is illustrative only.) ............... 26

Chapter 3

Figure 3.1. Gas puriﬁcation system connected to mass
spectrometer. Valves are configured such that during
purification, He bypasses the cold trap and travels onto the
GC column, while sample passes into the cold trap (N2) or
through the cold trap and then to vent via the mass flow
controller (N20). During analysis, helium passes through
the cold trap and onto the GC column, while the sample
vessel remains open to vacuum via the mass flow
controller. During N2 analysis the sample trap is not
chilled. Different columns for N2 and N20 are used.
Interface helium is used only for N2 analysis. The Penning
valve passes about 10% of the gas stream to the

spectrometer. ......................................................................................... 62

xviii

Figure 3.2. Measured vs. calculated molecular fraction of 45N20 for
moderate enrichments of N20, prepared volumetrically in
the laboratory from enriched and unenriched standards;
analyzed with the unmodiﬁed head ampliﬁer ................................... 64
Figure 3.3. Isotopic character of nine N20 samples from
laboratory denitrification of prepared ‘5N03' label, plotted
against the equilibrium curve. As expected, these samples
are in equilibrium. Vertical bars intersect the curve at
predicted enrichments, open circles represent measured
values ....................................................................................................... 66
Figure 3.4. N; and N20 fluxes with cumulative precipitation for
nine field incubations over the same highly labeled plot ............... 68
Figure 3.5. Estimated soil enrichment, comparison of MS and
GC, and N20 mole fraction for nine incubations
represented in Figure 3.4 ..................................................................... 70
Chapter 4
Figure 4.1. Change in headspace concentrations
(volume/volume) of nitrous oxide during incubation, with

selected (for clarity) standard error bars. Dashed lines =

successional soils, solid lines cropped soils. Open
symbols = control, ﬁlled symbols = pre-wet. Gray squares
= + acetylene, black diamonds = - acetylene. Control and

pre-wet curves diverge strongly for cropped soils (solid

lines with open or ﬁlled diamonds) but not for successional
soils (dashed lines with open or filled diamonds). ........................ 99
Figure 4.2. Summary of production of N2 and N20. “Total
denitrification" is N20 + N2 by acetylene inhibition, “N20” is
production of N20 in the absence of acetylene, and “N2" is
the difference. Even though total denitriﬁcation did not differ
significantly between moisture histories for cropped soils
(bars with solid borders), the control incubations produced
significantly more N20 and correspondingly less N2. Bars
with dashed borders represent soils from successional
plots ........................................................................................................ 101
Figure 4.3. Isotopic character of N2 for labeled and unlabeled jars
at the end of the incubation, paired (by line segment) to
represent incubation units. Dashed lines = successional
soils, solid lines = cropped soils. Open symbols =
controls, ﬁlled symbols = pre-wet. Circles = Replicate 4,
squares = Replicate 5, triangles = Replicate 6. Due to the
overwhelming abundance of unlabeled N2 from the
atmosphere relative to labeled soil—derived N2,
displacement of isotopic character during the incubation
(i.e., length of the line segments) is much smaller than for
N20 (Figure 4.4). Only a small portion of the equilibrium

curve (downward-opening parabola in Figure 4.4) is visible

at this scale. For cropped soils, control segments (solid
lines with open symbols) are much shorter than pre-wet
segments (solid lines with closed symbols) indicating less
N2 production for controls. For successional soils (dashed
lines) the opposite effect or no effect is seen. ............................... 103
Figure 4.4. Isotopic character of N20 for labeled and unlabeled
jars at the end of the incubation, paired (by line segment) to
represent incubation units. Dashed lines = successional
soils, solid lines = cropped soils. Open symbols =
controls, ﬁlled symbols = pre-wet. Circles = Replicate 4,
squares = Replicate 5, triangles = Replicate 6. Due to the
overwhelming abundance of N20 from soil relative to
unlabeled atmospheric N20, enriched samples represent
essentially the isotopic character of N20 derived from soil.
Samples falling on or near the equilibrium curve
(downward-opening parabola) indicate a source in isotopic
equilibrium, implying a single, uniformly labeled substrate
pool (eg. a homogenous mixture of native NO; and ”NO;
label). ..................................................................................................... 105
Figure 4.5. N2 production calculated independently by the
acetylene inhibition technique (AIT) and by mass

spectrometry (MS), for both cropped and successional

soils. In every case but one, MS gave a much smaller

estimate of N2 production than did AIT. ........................................... 107

INTRODUCTION

Denitriﬁcation in soil is the major source of atmospheric nitrous oxide
(N20), a potent contributor to global warming and to the destruction of
stratospheric ozone. Flux of N20 to the atmosphere is poorly understood.
There is a substantial imbalance in the global N20 budget (missing sources),
and N20 ﬂux at the ﬁeld scale is difﬁcult to predict, even when the rate of
denitriﬁcation has been adequately characterized. Much of the uncertainty
regarding N20 ﬂuxes from soil may arise from the highly variable relative
proportions of N20 and N2 produced during denitriﬁcation, commonly
expressed as the N20 mole fraction. The primary objective of my dissertation
research was to explore the influence of various ecological factors on N20 mole
fraction. A supporting objective was to develop robust methods for
characterizing N20 mole fraction in the laboratory and ﬁeld.

The structure of this dissertation reﬂects the objectives listed above.
Chapter 1 reviews current knowledge regarding factors that influence N20 mole
fraction directly and indirectly. Chapter 2 is reprinted from a publication in
which we present theory and equations that facilitate the interpretation of 15N
data for N-gas ﬂuxes from soil. Chapter 3 explains our development of
methods for measuring N-gas ﬂux and the N20 mole ratio. We combine
chamber methods and mass spectrometry in a way that allows very sensitive
measurement of N2 and N20 ﬂux from 15N labeled soil. The goals were to

improve on the statistical uncertainty of traditional methods that require

separate chambers for N2 and N20 measurements and to mitigate the inherent
bias of 15N methods by using identical methods for N2 and N20 when
estimating ratios. Chapter 4 deploys both traditional and isotope methods to
examine an empirical question: what are the effects of short term moisture
history and ecosystem management history on N20 mole fraction? Though
many questions remain, I am convinced that we have made signiﬁcant

progress.

Chapter 1

FACTORS INFLUENCING THE RELATIONSHIP BETWEEN DENITRIFICATION
AND NITROUS OXIDE FLUX FROM SOIL.

Background

Human domination of ecosystems and biogeochemical cycles has
become widely recognized (Vitcusek et al. 1997). Anthropogenic changes in
ecosystems can generate changes in the Earth’s atmosphere, which in turn
can have widespread consequences. Nitrous oxide is one of several
atmospheric trace gases that have attracted the attention of the scientiﬁc
community. It is produced in soils by microbial denitrification (Robertson
1999), nitriﬁcation (Firestone and Davidson 1989) and other biological
processes (Robertson and Tiedje 1987). The concentration of N20 in the
Earth’s atmosphere is currently about 312 ppb, (= 312 - 10'9 L - L") and is
increasing at a rate of 0.5 ppb per year (IPCC 1996). Pre-industrial levels were
< 290 ppb, suggesting that current levels reflect a signiﬁcant anthropogenic
inﬂuence. Nitrous oxide has a long (~120 y) half-life in the stratosphere. lt
indirectly catalyzes the destruction of stratospheric ozone, increasing the
hazard of UV exposure at the Earth’s surface (Hahn and Crutzen 1982). Nitrous
oxide is also a radiatively active gas; about 5% of the observed radiative forcing
of climate change over the last 100 years can be attributed to N20 (Bouwman

1990)

International concern over rising levels of atmospheric nitrous oxide has
stimulated research regarding its global sources and sinks. An estimated 14
T9 NZO-N are consumed or stored in the atmosphere annually, while only
about 7.9 Tg of sources have been identiﬁed at the Earth’s surface (Davidson
1991). A large part of the imbalance in the nitrous oxide budget may be due to
uncertainty in ﬂux estimates, especially for tropical regions of the world. In
particular, contributions from agriculture may be underestimated (Mosier et al.
1998, Kroeze et al. 1999).

Most of the increase in nitrous oxide flux to the atmosphere may be
related to land use change, especially conversion to agriculture. Deforestation
produces pulses of nitrous oxide, whether by clear-cutting (Robertson and
Tiedje 1988) or burning (Davidson 1991). Nearly 100 T9 of anthropogenic
fertilizer N is applied to crops globally each year (Eichner 1994). Most of the
anthropogenic N applied to watersheds in the North Atlantic drainage basin is
eventually denitriﬁed (Howarth et al. 1996). If even a small fraction of fertilizer N
results in N20 ﬂux, contribution to the global budget could be signiﬁcant.
Eichner (1994) estimates that as much as 2.1 T9 N20-N - y'1 may derive from
fertilizer, similarly Matthews (1994) estimates the fertilizer source at 2.0 T9 N20-
N - y". Mosier et al. (1998) place emission of NZO-N from agricultural soils for
1989 at 2.1 T9.

Challenges for research
Although denitrification is well understood, the relationship between

denitriﬁcation and N20 flux from soil is not simple. First, other processes,

especially nitriﬁcation, produce N20 in soil (Firestone and Davidson 1989).
While nitriﬁcation dominates in well-aerated soils and denitriﬁcation dominates
under anaerobic conditions, both can contribute simultaneously to N20 ﬂux in
soils of intermediate aeration (e.g. Stevens et al. 1997, Panek et al. 2000), or
even low aeration (Wolf and Russow 2000). The relationship between soil
moisture and relative importance of nitriﬁcation and denitriﬁcation for N20
production is uncertain (Davidson 1991). Over large temporal and spatial
scales, flux of N20 from nitriﬁcation is smaller and more predictable, while
denitriﬁcation accounts for the bulk of N20 ﬂux but is highly variable temporally
and spatially.

Second, denitriﬁcation has another major product: dinitrogen.
Dinitrogen, unlike nitrous oxide, is formed only by denitriﬁcation, and
represents closure of the nitrogen cycle. The relative proportions of N20 and N2
produced during denitriﬁcation vary widely. Field studies have shown the
inadequacy of using a simple conversion factor to estimate N20 ﬂux from
denitriﬁcation (Vinther 1984) or denitriﬁcation from N20 ﬂux (Weier et al. 1993).
The N20 mole fraction during denitriﬁcation (i.e., N20l[N20 + N2] ) can vary from
zero to one (e.g. Rolston et al. 1982), is influenced by a suite of physical and
chemical factors (Firestone and Davidson 1989), and is unstable over time
(Letey et al. 1980a).

Third, N20 ﬂux during denitrification in soil is partly uncoupled from N20
production per se by factors which regulate the transfer of N gases from soil

solution to the atmosphere. N20 may be released from soil hours or days after

it is formed, while N2 produced at the same time may diffuse away from the
source more quickly due to lower solubility in soil water (Letey et al. 1980b).
Other factors, such as freeze/thaw cycles and sudden changes in barometric
pressure, may also modulate N20 ﬂux independently of N20 production.

Fourth, measurement of denitrification is difficult. Chamber methods
and gas chromatography sufﬁce for estimates of N20 ﬂux. However, changes
in headspace concentration of N2 are nearly impossible to detect against the
high atmospheric background. Two methods of N2 measurement have been
recommended: acetylene inhibition and 1SN dilution (Mosier and Klemedtsson
1994). Both have disadvantages, which vary in importance depending on the
speciﬁcs of implementation.

This dissertation focuses on the second and fourth of the challenges
listed above: factors inﬂuencing N20 mole fraction and methods for
measurement of denitrification. N20 mole fraction is briefly reviewed below.
The two chapters following explore theory and methods, respectively, for
measuring denitriﬁcation products. A ﬁnal chapter advances understanding of
the relationship between denitriﬁcation and N20 ﬂux by testing for ecosystem
differences and effects of soil moisture history on N20 mole fraction.

Factors inﬂuencing N20 mole fraction during denitriﬁcation
Theory
Denitrification is an anaerobic microbial process that converts nitrate to

gases (Paul and Clark 1996) by means of the cellular enzymes nitrate

reductase (NR), nitrite reductase (NiR), nitric oxide reductase (NOR), and

nitrous oxide reductase (NOS). It can be summarized as follows:

 

2[ N03'15-R—> 2[ N02'1-’&> 2N0 ﬂ» N20 N°S > N2

Nitrate serves as an electron acceptor, most commonly for the anaerobic
respiration of organic matter. Almost all denitrifying organisms use 02
preferentially, with the result that denitrification is restricted to anoxic sites (Paul
and Clark 1996). Genera with denitriﬁers are found among the organotrophs,
chemolithotrophs, photolithotrophs, diazotrophs, thermophiles, and archaea,
although organotrophs are the principal denitriﬁers in soil.

In the denitriﬁcation sequence (above), NO and N20 are free
intermediates that may or may not be further reduced, depending on
conditions. Proximally, N20 production depends on the relative status of NOR
and N08. lf N08 is fully active, denitriﬁcation may result in NZ only; however, if
NOS is inhibited relative to NOR activity, N20 is produced. It is convenient to
express the relationship between N20 and N2 production as N20 mole fraction:
N20 / [N20 + N2]. Alternative formulations of mole fraction giving essentially the
same information appear in the literature (9.9. N; / N20 etc.).

A suite of proximal controls on N20 mole fraction have been consistently
identiﬁed, with few variations (Colbourn and Dowdell 1984, Sahrawat and
Keeney 1986, Firestone and Davidson 1989, Arah and Smith 1990, Bouwman
1990, Aulakh et al. 1992, Hutchinson and Davidson 1993). Table 1.1 lists

factors hypothesized or demonstrated to influence N20 mole fraction. Not

surprisingly, most of the factors that inﬂuence N20 mole fraction also inﬂuence
the overall rate of denitriﬁcation (e.g., Paul and Clark 1996).

By what mechanisms do soil physical and chemical factors influence
N20 mole fraction? Change in N20 mole fraction at the cellular level results
from a change in the relative rates of production and consumption of N20.
Environmental factors can change N20 mole fraction by altering the effective
availability of substrates, by altering the relative production and maintenance of
enzymes, or by inhibiting the function of enzymes. Betlach and Tiedje (1981)
demonstrated the importance of substrate availability. They manipulated soil
characteristics and followed each pool in the reductive sequence using 13N
label. They found that their results could be predicted qualitatively using a
simple Michaelis-Menten model for each reductive step (cf. Dendooven et al.
1994). Experimentation with the model system led to the generalization that
any change that slows the overall rate of denitriﬁcation leads to an
accumulation of N20. Additionally, they concluded that this effect alone was
sufﬁcient to account for their experimental results; differential inhibition of
enzymes need not be invoked. The Betlach and Tiedje paradigm is consistent
with most other experimental ﬁndings.

An important implication of the Betlach and Tiedje paradigm is that, for
many controlling factors, inﬂuence on denitriﬁcation rate and inﬂuence on N20
mole fraction are twin aspects of a single phenomenon: general modiﬁcation
of reaction rates. However, conceptual independence is warranted because

there are important exceptions (eg. the effect of nitrate concentration on N20

mole fraction). The hole—in-the-pipe model (Firestone and Davidson 1989,
Davidson 1991) maintains a functional distinction between denitriﬁcation and
N20 mole fraction as controls on N20 ﬂux. According to the model, N20 ﬂux is
subject to controls at 3 levels: factors inﬂuencing overall rates of nitriﬁcation
and denitriﬁcation (ﬂow through the pipes), factors inﬂuencing N20 mole
fraction (the size of the holes in the pipes) and factors inﬂuencing transport of
gases out of the soil. Figure 1.1 shows explicit relationships among these
three levels of inﬂuence and the proximal factors listed in Table 1.1. Notice that
if a factor has opposite effects on rate and N20 mole fraction, changes in N20
ﬂux tend to be buffered. Sahrawat and Keeney (1986) provide a rather
comprehensive review of these factors. Below, their work is summarized, with
supplemental material from other sources.
Temperature

Temperature generally has a positive effect on total denitriﬁcation and a
negative effect on N20 mole fraction. Some have questioned the importance of
temperature for predicting N20 mole fraction (Lensi and Chalamet 1982).
Temperature has a general inﬂuence on a wide range of biochemical
processes; its inﬂuence is likely to be complicated (Sahrawat and Keeney
1986).

Available carbon

Organic carbon added to soils can stimulate denitrification and

decrease N20 mole fraction (Sahrawat and Keeney 1986). Greater supply of

electron donors may create a demand for electron acceptors, causing N20 to

be reduced. This is consistent with early work by Nommik (1956). Weier et al
(1993) reported a decrease in N20 mole fraction with increasing carbon
availability. However, Dendooven et al. (1996a) discovered an increase in N20
mole fraction with added glucose.
pH

Soil pH is an important regulator of enzyme activity. Early work showed
an increase in N20 mole fraction at pH < 6 (leer and Delwiche 1954, Nommik
1956); this finding was supported by later research (see Sahrawat and Keeney
1986, and references therein). Focht (1974) modeled N20 flux from soils using
Nommik’s data: he incorporated an explicit effect of pH on N20 mole fraction.
Blackmer and Bremner (1978) disputed Focht’s model; they found that pH had
no effect in the absence of nitrate. Whereas leer and Delwiche (1954) had
suggested that low pH inhibits the reduction of N20 to N2, Blackmer and
Bremner (1978) claimed rather that nitrate inhibits the activity of nitrous oxide
reductase although it stimulates its production (cf. Blackmer and Bremner
1979) and that low pH interacts with nitrate to increase its inhibitory effect.
Koskinen and Keeney (1982) attribute the pH effect on N20 mole fraction to
either the greater sensitivity of N20 reductase to pH or to changes in species
diversity at low pH. They note (sensu Blackmer and Bremner 1978) that there is
no pH effect at low nitrate concentrations. They suggest that low pH may inhibit
NOS indirectly by favoring high nitrite concentrations. Weier and Gilliam (1986)

noted large ﬂuxes of N20 from low pH soils; ﬂuxes were correlated with

10

accumulation of N021 In general, pH control of N20 mole fraction has not been
demonstrated in the ﬁeld (Davidson 1991).
Nitrate and nitrite

It is universally reported that high nitrate concentrations increase N20
mole fraction (Nommik 1956, Cooper and Smith 1963, Blackmer and Bremner
1978, Vinther 1984, Weier et al. 1993). Firestone et al. (1979) found, however,
that nitrite was a more potent inhibitor of nitrous oxide reduction than nitrate (cs.
Gaskell et al. 1981, cited in Coulboum and Dowdell 1984), and suggested that
small amounts of nitrite produced from large pools of nitrate may in fact be
responsible for the inhibition usually observed.

Since nitrate, when limiting (Sahrahwat and Keeney 1986) is positively
correlated with denitriﬁcation and yet increases N20 mole fraction at high
concentrations, it does not conform in a strict sense to the Betlach and Tiedje
model. This suggests that its inﬂuence is not via simple substrate kinetics.
Probably nitrate concentration is not high and limiting at the same time; nitrate
may have a small but unreported kinetic effect on N20 mole fraction at low
concentrations but must exert inﬂuence by a different mechanism at high
concentrations. Two proposed mechanisms are that nitrate competes with
N20 as an electron acceptor or that it directly or indirectly inhibits NOS.
Firestone et al. (1979) favored the ﬁrst mechanism for nitrate but the second for
nitrite. This question has not been resolved, although the fact that pH interacts

with nitrate concentration suggests the priority of enzyme inhibition.

11

Soil Water and Oxygen

Three mechanisms are responsible for the inﬂuence of soil water on
N20 mole fraction. First, at very low soil moistures, metabolic stress may limit
denitriﬁcation (Sahrawat and Keeney 1986) and N20 mole fraction may be
inﬂuenced concomitantly. Second, at high soil moistures, Oz availability in soil
microsites is limited by diffusional transport of oxygen; denitriﬁcation and N20
mole fraction are affected. Third, high soil water content limits the diffusion of N
gases away from the sites of formation, increasing the probability of further
reduction of NZO-N. The ﬁrst two mechanisms are consistent with the kinetic
model of denitriﬁcation (Betlach and Tiedje 1981) and all three proceed in the
same direction with respect to the effect of soil moisture on N20 mole fraction.

Water-ﬁlled pore space, when expressed as a percentage of the total
pore space, is the most informative way of describing soil moisture because it
integrates absolute water content (eg. gravimetric) with bulk density (Linn and
Doran 1984). It is probably the single most useful predictor of N20 mole
fraction, largely because of its inverse relationship with soil aeration.
Denitrification is anaerobic, and all denitriﬁcation enzymes are sensitive to
(inhibited by) oxygen.

Lower soil water contents favor N20 production over N2 (Nommik 1956,
Weier et al. 1993). However, the effect of water status is inseparable from its
effect on soil aeration. Oxygen inhibits denitriﬁcation but increases N20 mole

fraction (Firestone et al. 1979). Focht (1974) included aeration as a

12

determinant of N20 mole fraction in his model of denitrification. Implications for
enzyme status are discussed below.
Enzyme status

The principal effect of oxygen is to inhibit denitriﬁcation enzymes. The
most sensitive of these is the terminal enzyme NOS. As oxygen increases,
reduction of N20 to N2 slows sooner than reduction of NO to N20, thereby
increasing N20 mole fraction (McKenney et al. 1994). Notwithstanding, the
Betlach and Tiedje model predicts increasing N20 mole fraction with
increasing Oz, regardless of enzyme sensitivity. If a well-aerated soil is
sufﬁciently wetted that it becomes anaerobic, denitriﬁcation enzymes are
induced sequentially (see Letey et al. 1980a). Thus some N20 accumulates
before NOS is completely induced, and N20 mole fraction should decrease
with time. The ever-present potential for synthesis and inactivation of enzymes
makes N20 mole fraction inherently dynamic. Enzymes may be induced,
repressed, de-repressed, re-repressed, or destroyed. Any difference in the
time-dependent behavior of the last two reduction enzymes leads to a time
dependent difference in accumulation of the last two reductive products - N20
and N2. Just as water status is usually inseparable from Oz status, enzyme
status is inseparable from its temporal dependency.

Time

Changes in N20 mole fraction may lag behind changes in factors that

control it, creating complex temporal dependency. Rolston et al. (1978) found

that N20 mole fraction was highest at the initiation of denitriﬁcation, and

13

decreased thereafter. Jacinthe et al. (2000) manipulated water tables in soil
columns (various drainage classes) to stimulate denitrification; N20 mole
fraction was 0.95 four days after raising a water table to 10 cm below the soil
surface, but dropped to 0.35 one week later. Firestone and Tiedje (1979)
reported that after the onset of anaerobiosis, N20 mole fraction was initially low,
then rose signiﬁcantly, and then declined. They attributed this effect to the
staggered synthesis of enzymes, and speculated that it reﬂected the sequence
of events following 02 depletion due to rainfall or irrigation. Letey et al. (1980a)
reported high initial values for N20 mole fraction during denitrification, with a
subsequent decrease to zero. They attributed this effect to differential rates of
induction of nitrate reductase and nitrous oxide reductase. Rolston et al.
(1982) found that “nitrous oxide mole fractions tended to be smallest
immediately after irrigation and increased as the soil water redistributed and
became less anoxic”. They also found a decrease in N20 mole fraction with
successive irrigation cycles. Hallmark and Terry (1985) reported that N20 mole
fraction following irrigation was initially high, then dropped over a period of six
weeks. Weier et al. (1993) reported N2/N20 ratios rising (i.e., N20 mole fraction
dropping) over a 5-day period. Dendooven and Anderson (1994) found that de
novo synthesis of nitrate reductase and nitrous oxide reductase began one and
16 hours after anaerobiosis was imposed, respectively. Dendooven and
Anderson (1995) found that upon return of soil to aerobic conditions, N20 mole
fraction increased with time over a 70 day period (from 51% to 100%).

Conversely, soil cores submerged for 96 h had smaller subsequent N20 mole

14

fraction than soils submerged for 6 h (Dendooven et al. 1996b). In general, it is
difﬁcult to predict N20 mole fraction without knowing both the antecedent water
regime and the time since change in water regime.

Effects of moisture cycles on denitriﬁcation and N20 mole fraction merit
consideration. Arnold (1954) suggested that moisture fluctuation should
stimulate gaseous N loss from soils. High denitriﬁcation rates are sometimes
attributed to extreme drying-wetting cycles (Peterjohn and Schlesinger 1991).
Smith and Patrick (1983) found that 7 and 14 day aerobic/anaerobic cycles
produced far more N20 than either condition alone. A simple explanation is
that the cycling systems represent intermediate levels of aeration, such that
denitriﬁcation is supported, but not complete reduction to N2. However,
moisture cycles may support more complex dynamics. Groffman and Tiedje
(1988) showed that hysteresis (dependence of the response on the direction of
change) is important for the overall rate of denitrification. Hutch et al. (1999)
found N20 emission under a fluctuating moisture treatment 4 to 9 times higher
than under constant low or high moisture. The ﬁnal chapter of this dissertation
describes an investigation of the temporal dependence of N20 mole fraction on
soil moisture, i.e., the effect of moisture history.

Prospectus

The task of understanding the relationship between N20 flux and
denitriﬁcation will have been accomplished when N20 from denitriﬁcation can
be predicted reasonably well from a discrete set of measurable factors. Soil

water status will likely continue to be a focus of investigation because it is

15

easily measurable and correlates well with oxygen status. Oxygen status,
though more difficult to measure in spatially heterogeneous soil environments,
is the most important proximal regulator of denitriﬁcation enzyme status.
Dynamic, differential variations in enzyme status are the principal cause of
uncertainty regarding the relationship between N20 ﬂux and denitriﬁcation
because of the potential for production of the alternative product N2. From an
empirical perspective, useful approaches will be (1) to investigate the
response of N20 and N2 ﬂux to differences in soil moisture (2) to assess the
temporal dependence of the relationship between ﬂux and moisture, and (3) to
test for the robustness of the perceived patterns across varying environments.
In the work that follows, I seek to implement these approaches by developing
techniques for observing N20 and N2 flux, and by testing explicitly the temporal
dependence of ﬂux on change in moisture, for soils from contrasting

ecosystems.

16

Table 1.1. Summary of factors potentially inﬂuencing N20 mole fraction. The

generally-accepted direction of the correlation is indicated. WFPS = water-ﬁlled

 

pore space.

factor correlation
temperature (0 - 30 °C) -
water (60-100% WFPS) -
Time since water addition -
nitrate, nitrite +
available organic carbon -
pH (4 - 7) -
902 “I“
enzyme status variable
rhizospherel plants ?
depth of activity -
soil structure +

17

Figure 1.1. Conceptual diagram: factors inﬂuencing ﬂux of N20 from soils.
WFPS = water-ﬁlled pore space. C = organic carbon availability. P =

phosphorus availability.

18

. F. F 2:9“.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   
    
 
  

 

 

 

 

 

\_. SeaEzcocoma 208 ONZ _
l +
new . -
snags. his E E E .I
Cm «MZONZ l
a/TLI- 3 z..+ + + -+«
+\®\ + SEEEZ 30 can
Beaumont“
a use onz
+/$/ \_ gaﬁécouoﬁ 22a ouz
l +
mam - w +
- 3.52... a . E B -E E
s 302 a- -
+
BESEEQ a cam

 

 

era. #3

E5

19

Chapter 2

A HEURISTIC MODEL FOR THE CALCULATION OF DINITROGEN AND
NITROUS OXIDE FLUX FROM NITROGEN-15-LABELED SOIL

Published in
Soil Science Society of America Journal 63: 1709-1716 (1999).

20

Wham theSoilSeiaccSna‘eoro/mm
Volume 63. no. 6. Nam-Dec. 1999
6775m5eguM.MadimV/153711USA

A Heuristic Model for the Calculation of Dinitogen and Nitrous Oxide Flux
from Nitrogen-IS-Laheled Soil

Timothy T. Bergsma.‘ Qiaobing C. Bergsma. Nathaniel E. Ostrom. and G. Philip Robertson

ABSTRACT

VeryseasitivematsongaadNﬁﬂufronsoilan
Mute-gandvedfromuN-lﬂeledsoilisaaalyzedbyisotope
ratiomspectronetry.1'hisapproachisusefulforstudyiagthe
Hedaiuugenlertilizerandforstudyingsoilmicrobialproceases
coatriiuting to the atmospheric increase ofnitrotnoxideJradiatively
activetracegasthatcanconuibutetoyobalwamingandoaoae
depletion. Most system of equations that relate isotopic analysis to
psﬂuxaresulﬁciently complexthatcerta’n limitatiousaadpoteatials
olthe ”N approach may be overlooked. We describe a graphical
representationollabeledN-gaslluxthatMeatheequatioas-d
Woitialth‘nkingregardingtheinplemeatationdrelated
eaperineatsﬂismodelisusedtointerpretanderestiatatioathat
msifhxderivesfromnalt’quepoolsofdiﬂeringenrichmenA
statistialderivationispreaented lorapreviously published simulation
of-derast'naﬁoadueto-ultiplepoohnesaneequatioasare
qphed to field data to explore whether temporal variation in soil
ﬁateearichneatislihelytocausesigﬂﬁcant nderestinaﬁomTwo
“WmWM-ayehhatethempﬁoa
otpoolﬂordttherebyeiaa’aatiagapoteatialmoluderesti-

nation.
D lNlTROGEN AND NITROUS oxror: are alternative end
products of microbial denitriﬁcation. Quantifying
their flux from soil can help explain fertilizer losses from
agricultural systems (Mosier et al.. 1986: Eichner. 1990'.
Weier et al.. 1993) as well as the atmospheric buildup
of N20 — an important greenhouse gas in the tropo
sphere and ozone-destructive catalyst in the strato-
sphere (Bouwman. 1990: lPCC. 1996). However. ﬁeld
studies of denitrification have been hampered by the
insensitivity of standard instrumentation to N2 increases.
e.g., under soil covers (Mosier and Klemedtsson. 1994).
Mass spectrometric analysis of gas from 'SN-labeled
soil is a sensitive method for quantifying the flux of N2.
as well as NO. because of the low natural abundance
of '5N. Isotopic data for headspace gases can also esti-
mate the enrichment of the source N pool. can help
identify the source of N for N20. and can be used to
check assumptions of the ﬂux method (e.g., Stevens et
al.. 1997 : Arah. 1997). Recent advances in spectrometer
sensitivity and affordability have generated new interest
in using "N for soil biogeochemical investigations.
Given the complexity of the system of equations nor-
mally used for interpreting the isotope data. it is useful

 

T.T. Bergsma and G.P. Robertson. W.l(. Kellogg Biological Station
and Dep. of Crop and Soil Sciences. Michigan State Univ.. Hickory
Corners. Ml 49060. QC. Bergsma. MedFocus Clinical Research Con~
sulting Services. Des Plaines. ll. “18: NE. Ostrom. Dep. of Geologi-
cal Sciences. Michigan State Univ.. East Lansing. MI 48823. Received
16 Oct 1998. ‘Corresponding author (tbergsma@kbs.msu.edu).

Published in Soil Sci. Soc. Am. J. 63:1709-1716 (1999).

1709

21

to have conceptual tools which make the principles of
the system more intuitive.

We present here a heuristic model that illustrates the
estimation of source pool enrichment and estimation
of soil-derived headspace gas for N; or NO from 1’N-
labeled soil. A convenient graphical representation of
N isotope data is identified. from which a complete set
of equations is derived by geometric inference. These
equations are similar in form and identical in function
to those of others (Siegel et al.. 1982; Mulvaney. 1984;
Arah. 1992). To illustrate utility, the model is applied
to the problem of underestimation that occurs when
ﬂux derives from multiple pools of differing enrichment.

Measuring ﬂux

Flux of N2 produced by denitrification in a "N-
enriched soil can be measured by monitoring the in-
crease in enrichment of headspace gas in a chamber
placed over the soil. Proper analysis requires measure-
ment of the abundance of all three molecular masses
of N3 (28, 29. 30). When atoms of lsN are distributed
randomly among a sample of N2 molecules. measure-
ment of any two masses sufﬁces. because the abundance
of the third mass can be predicted statistically. However.
a mixture of N; from two differently labeled sources
(e.g.. enriched N; from the soil and unenriched atmo-
spheric N2 in an enclosure) is not in isotopic equilibrium
(Hauck et al.. 1958); that is. the isotopes of N are not
randomly distributed among the three molecular frac-
tions. This means that all three masses must be mea-
sured. which has the additional advantage of providing
an indirect estimate of the average enrichment of the
soil N pool (Hauck and Bouldin, 1961). The estimation
of enrichment is convenient because it is non-destructive
and is a time-weighted mean.

Equations for the determinations of ﬂux and source
enrichment, by isotope ratio mass spectrometry. are well
established (Siegel et al., 1982; Mulvaney and Boast,
1986; Mulvaney, 1984; Arah. 1992). These equations are
designed to measure total ﬂux of N2 and assume that
the gas is derived from a single, unifomly labeled pool
of soil N. (In the absence of pool uniformity. flux derived
from added label—cg, fertilizer—can be estimated. but
not by these equations.) The same principles apply for
N30; for simplicity, most of this discussion is limited
to N2.

Graplical Representation
We adopt, wherever possible, the notations and defi-
nitions of Arah (1992). Brieﬂy. the "N atom fraction l’a
of any sample of N is the total number of l5N atoms

1710 son. scr. soc. AM. 1. VOL 63. NOVEMBER-DECEMBER 1999

divided by the total number of N atoms (informally. the
enrichment). The molecular fractions 21‘x. 2"x. and ”x are
the fractions of the total number of N; molecules in a
sample with masses 28. 29. and 30. respectively. A sam-
ple of N3 is in isoropic equilibrium if the molecular frac-
tions follow a binomial distribution:

28x = (Na)! [la]
2"x = 2(“a) (‘50) [1b]
-‘°x = ("of [1c]

where
" = l - 15a [1d]

All N: (or NO) derived from uniform pools of mole-
cules with single N atoms (e.g., nitrate) by microbial and
chemical processes (e.g., denitriﬁcation) are expected to
be in isotopic equilibrium. even if isotopic fractionation
is signiﬁcant (i.e., even if one isotope is inherently fa-
vored by the process). All mixtures of two or more
equilibrium samples with different "N atom fractions
are necessarily nonequilibrium mixtures. Atmospheric
N2 is assumed to be in isotopic equilibrium.

Additionally, we define the isotopic character of an N;
sample as the relative proportions of the three molecular
fractions. We represent the isotopic character of N; by
plotting ”1 versus ”a. The fact that such a plot is equiva-
lent to a ternary plot of ”x, ”x. and ”r: (Fig. 1; Note 1)
is both a mathematical convenience and an informal
proof that ”x and "a completely characterize N2 isotopi-
cally. The isotopic character of a single sample may be
expressed as the Coordinate pair (”a. ”x ).

Figure 2 represents graphically the calculation of
source pool enrichment and flux from a plot of isotopic
character. The downward-opening parabola is called the
equilibrium curve. and has the form of Eq. [1b]; it repre-
sents the isotopic characters of all possible N2 samples
that are in isotopic equilibrium. Whenever N atoms are
paired randomly from a source of given enrichment.
”N3 will be a quadratic function of IN. with no ”N,
produced when all or none of the source atoms have
mass 15.

The symbol A represents the isotopic character of
atmospheric N3 already present in a chamber headspace.
P represents the isotopic character of N2 derived from
the soil pool: since the soil pool consists of uniformly
enriched mineral N. Component P is initially in isotopic
equilibrium. M represents the isotopic character of a
mixture of atmospheric and soil-derived N2 in a chamber
headspace (this mixture is not in isotopic equilibrium).
Plotted thus. the isotOpic character of any mixture is a
linear interpolation between its two constituents. and
its distance from either is inversely proportional to the
relative contribution from that constituent. Thus M falls
on a line between A and P. and its position along that
line indicates the mixing ratio of A and P.

In practice. a soil cover is deployed. and gas samples
are taken at the beginning and end of an incubation. A
and M are the isotopic compositions of these samples.
respectively. P is unknown initially. but must fall some-
where on the ray drawn from A through M. and must

22

also fall on the equilibrium curve. Thus. the intersection
of the curve and the ray identifies the character of P.
The atom fraction of the soil N pool, identical (barring
fractionation) to that of P. is displayed on the horizontal
axis. The relative contribution of P to the mix is given
by the “travel” of M along the ray (the length of segment
AM). divided by the total length of the segment AP.
Since the ray has constant slope. the relative contribu-
tion can be determined simply from the relative enrich-
ments ("N) of A. M. and P (i.e., the proportion collapses
to a single axis).

Equations

From the concept illustrated in Fig. 2. equations for
gas flux and enrichment of source pool can easily be
derived by geometric inference. Formally, the isotopic
character of a gas mixture and the assumed or measured
character of one of its pre-mixing components (e.g.,
atmospheric N2 already present in a chamber head-
space) can be used to calculate both “a,. the atom frac-
tion of the second component (e.g., soil-derived N2),
and d, the fractional contribution of the second compo-
nent to the mixture. This is achieved by assuming that
the second component is initially in isotopic equilibrium
(as stated earlier). The atom fraction of the second com-
ponent is found by simultaneous solution of an equation
for the mixing line

”x = C + 5% [2a]

and an equation for the equilibrium curve [1b], which
results in a quadratic expression:

"a, = {-3 z (132 - 4AC)"’] I 2A [2b]

where
A = 2 [2c]
8 = s - 2 [2d]
C = ”x. - 5050.) = ”x. - 3("0.) [2c]
= (29x- ._ 29x.)/(IS . _ 15“,) [2f]

and the Subscripts a. p, and m refer respectively to the
initial component (atmospheric N), the second compo-
nent (soil pool N). and the mixture. C is the intercept
for the mixing line, and s is the slope of the mixing
line. The relative contribution of N; from the second
component is

d = (15 _ ”ad/(”0, _ ”a.) [3]

Absolute flux can be calculated from relative flux (rela-
tive contribution) by associating some absolute measure
with initial or final quantity of headspace gas: for exam-
ple. if final N, concentration is 0.8. and chamber volume
is 1 L. then a value of 0.01 for d implies evolution of
(0.8)(1 L)(0.01) = 0.008 L of N2. Note that a quadratic
expression is used to find ”a, which in turn is used to
find d. Arah (1992) used a quadratic to find if and used
d to ﬁnd ‘5a,,. The two systems give identical results. The
relevant root in Eq. [2b] can be identified by inspection
of Fig. 2 and is necessarily the greater; the other root
is identical to a. in this case (cf., Boast et al., 1988 Eq.

BERGSMA ET AL: HEURIS‘nC MODEL or nmtooermxs FLUX mom NHROGEN-IS-LABELED SOIL 1711

 

"Na molecular traction x

 

 

“N atomic ﬁction I

Fig.1.MMMWMteruyaadCandmmtahwuon-dﬁalﬂedNxmmuﬁmm
independentiyrandomized,Waatern-ymthbeaesaneoletho-alduplesplottdas’xu‘n.

[26]. [31] ﬂ“. ). Note again that all mixtures of equilibrium never past it. The formal proof has been provided by

gases necessarily be under the equilibrium curve. This Boast et a1. (1988; see Eq. [19]).

is true regardless of the number of equilibrium compo.

nents in a mixture. As an informal proof. consider step— .

wise additions of many equilibrium components to an Enid?!” “12°: linorbDue

accumulating mixture. The ﬁrst two components create 0 p e 0°

a mixture below the curve; subsequent additions dis- The utility of the heuristic model outlined above can
place the mixture toward some point in the curve. but be illustrated by revisiting the principle assumption of

23

1712 SOIL scr. soc. AM. 1.. VOL 63. NOVEMBER—DECEMBER 1999

0.5 1'

 

P

0.4 <
x F
g 0.3J
a
3

0.2
3
ii

0.1 A? M

A .
o i 4. i 4. i t j

 

0 0.1 0.2 0.3 0.4

0.5 0.6 0.7 0.8 0.9 1

“Nannietacttona
Fig.2. Visaah‘aatioaotthehe-istie-odelfl‘lq-IILIZL-‘Dll-AWhmduwﬁeqlepa‘a-db‘
www.mmhnMAWMMhMmhﬂaﬁgadedeI-ﬁe

aim-mama“

disequilibrium methods Of N; or N20 flux analysis
(Hauck and Bouldin. 1961; Siegel et al., 1982; Mulvaney,
1984; Arab. 1992). Any method that does not measure
enrichment of the soil mineral N substrate directly. but
infers it from isotOpic data, assumes that soil N; or N20
derives from a single, uniformly labeled pool. The im-
portance Of this assumption has been debated (Focht,
1985; Mulvaney and Kurtz. 1985) and evaluated in the
lab (Mulvaney, 1988) and field (Mulvaney and Vanden
Heuvel. 1988) and tested by simulation (Vanden Heuvel
et al., 1988; Arab, 1992). Addition Of ”N-labeled mate-
rial to native soil pools creates the possibility Of at least
two differently labeled pools. Theory suggests that con-
current flow from multiple pools of different enrichment
should usually lead to an overestimation Of soil N enrich-
ment, and should always lead to an underestimation Of
flux (Boast et al.. 1988; Arah. 1992).

Arah (1992; see Fig. 2a) simulated the underestima-
tion that occurs when the assumption Of pool uniformity
fails. Estimated mixing ratio d was compared with actual
mixing ratio for 1000 runs in which number of pools
was randomized on the Interval [1, 50] (cf., Vanden
Heuvel et al.. 1988, for simulations based on two pools)
and pool enrichment was randomized on the Interval
[7, l] (where 'y is the natural abundance of "N in the
atmosphere. or 15a.). The plot Of estimated vs. actual
showed very little scatter about a line with a slope of
0.76; thus, for the conditions that were simulated, the
isotopic method consistently underestimated N2 flux by
about 24%. This non-intuitive result begs for a more
fundamental interpretation.

We have discovered that Arah’s result can be calcu-
lated directly from statistical principles, without re-
course tO simulation. It is instructive to begin by visualiz-
ing how underestimation occurs (Fig. 3). If a gas sample
P'“““ is itself a mixture Of N; that derives from two
or more soil N pools. then the character of P will fall
somewhere below the equilibrium curve (Fig. 3). The
ray approaching P”"" from the direction of A will neces-

24

sarily “overshoot” P, intersecting the curve beyond P.
Thus, I5a,, will always be overestimated if 15a, > "a.
(as shown by Boast et al., 1988; and Arah, 1992). It is
apparent from Fig. 3 (and from Eq. [3]) that d is always
underestimated: the segment AW is always longer
than the segment AP‘“"‘ (as per Arah, 1992).

The underestimation can be quantified if the isotopic
character of P‘““" is specified. Regarding Arah’s (1992)
simulation. it is appropriate to associate P“ with the
statistical expectations of "a and ”x for a mixture of
equilibrium gases. whose values of 1’a are randomly and
uniformly distributed on the interval [g, h]. With P“
thus speciﬁed, the expectation Of underestimation can
be calculated. The answer derived here is independent
of the number Of contributing gases (pools) because
statistical expectation E(x) is independent of the num-
ber Of samples. For generality, we take the case where
each pool has equal weight. For a random variable uni-
formly distributed on the interval [g, h].

5050) = (g + Ill/2 l4]
Var("a) = (h - g)’/12 [5]
E("a’) = E2(“a) + Var("a) [6]

Considering the dependence of ”x on "a (Eq. [1b]), the
expectation of ”x is given by
E(”x) = E(2 l5a - 2 "a1) [7a]
= ZE("a) - 2E("a’) [7b]
Substituting [5] in [6] and [6] in [7b] gives
E(”x) = ZE("a) - 2 [Ez("a) + (h — g)’l12] [8]
Thus, given a range of enrichments, the expected isoto-
pic character Of a random mixture PM, ”1““)
can be determined from [4] and [8]. To calculate the
resulting underestimation, an atom fraction "a,” is
found by substituting ”W and "am for ”x. and
l5a... in Eq. [2]. A coefﬁcient of estimation e can be
expressed as

BERGSMA ET AL: HEURISTIC MODEL OF NITROGEN-GAS FLUX FROM NI'IROGEN—IS-LABELED son. 1713

0.5 T

"N: molecular traction x
9 9 .o
N W .
‘ l 1r

O
.a
i
K

 

0 t ¢ ¢ ¢
0 0.1 0.2 0.3 0.4

A 1 4* 4
Y I V 1' ﬂ

0.5 0.6 0.7 0.8 0.9 1

"Natomictctiona
FglvmdmmmWMNMdMMHPihMa-n—ed

(redthglron-ltiplepoolsnhea
WAWBWMAF,M

e = doubted/dew!“ [9a]
= (1509mm _ ‘50,)/(”ap°’°““‘ _ 150.) [9b]

Eq. [9b] can be proven from Eq. [3]. For the case where
the range of enrichments is [0.1], ”mm = (V2) by
Eq. [4] and ”W = (1/3) by Eq. [8] (Note 2). Inter-
estingly. if the lower bound of the interval 3 is equal to

y(‘5a.: cf. Arah. 1992). e— = 0.75. regardless Of the range
Of enrichment. this value depends neither on the value
of 1 nor it (Note 3).

In summary, we have provided a more fundamental
interpretation of Arah‘s (1992) simulation (Fig. 2a). Our
heuristic model for N-gas ﬂux from lsN-labeled soil
shows why underestimation occurs. Our equations re-
duce the simulation to a relatively straight-forward cal-
culation. We show that even under somewhat less restric-
tive conditions than Arah’s (i.e., range not speciﬁed). the
coefﬁcient Of underestimation e evaluates to 0.75. This
value agrees well with the slope of 0.76 in Arah’s Fig.
2a, which can also be interpreted as an index Of underes-
timation. Simulation is a valuable tool for exploring
systems of equations that defy direct solution: the dis-
covery Of a direct solution for such a system represents
progress. While our solution may not represent any real
set of ﬁeld conditions, it does help predict how ﬁeld
conditions will influence the accuracy Of isotopic meth-
ods for N-gas flux measurement.

The approach employed above can be used to explore
other questions about labeled N; or NO ﬂuxes. For
instance, Hauck and Bouldin (1961) state that their sys-
tem gives a value for 15N that “represents the average
isotope content of the material undergoing denitriﬁca-
tion over a given period of time.” But the concept Of
“average” implies that the enrichment is changing with
time (e.g., by dilution from concurrent nitriﬁcation).
and thus the assumption of pool uniformity is violated.
In principle. it makes no difference whether the assump-
tion is applied to space or time. Is change Of source pool
enrichment with time likely to be a signiﬁcant source
of error in ﬁeld experiments?

25

RWMWMWmMPHMWF,m“g>'~Shaﬁe
isalwaysaaderestiaatedts‘ueelluxAMIAP).

We sampled N20 over a 3.5-d period in April 1998
in a heavily labeled wheat plot (30 kg ha" as 99%
1(”NO3). Enrichment of the source pool, as inferred
from N20 isotope data. dropped gradually from 82 to
72% during this period. Even if this entire drop had
occurred during a single incubation, the resulting under-
estimation would have been negligible. Let [g. h] be
[0.7, 0.8] and let g = am. From Eq. [9b] (which
invokes Others), e = 0.999. For comparison, a drop from
80 to 60% during an incubation yields e = 0.993. We
conclude that error from temporal changes in enrich-
ment for our experiment must have been negligible, and
is probably negligible in most cases.

DISCUSSION

The heuristic model presented above facilitates the
design of N-gas flux experiments and the interpretation
Of isotopic data for N2 and N20 samples collected over
”N-labeled soil. It is particularly useful for exploring
the problem of underestimation that occurs when N; or
N20 analyzed by mass spectrometry derives in part from
a soil pOOl that is not uniformly labeled: it illustrates
how underestimation occurs. We reduced a published
simulation of underestimation to a direct calculation
based on statistical principles. We showed the general
utility of our equations by evaluating a case where en-
richment varied over time, rather than in space.

Isotopic methods for measuring N2 and N20 flux have
general appeal because they are relatively non-disrup-
tive Of soil systems and because they represent the only
practical direct method for measuring N2. As isotope
ratio mass spectrometry (IRMS) becomes more widely
available, use of isotopic methods will continue to grow.
Although flux equations and analytical methods have
been available for decades, there still exists considerable
uncertainty regarding the accuracy of the method when
applied in the ﬁeld. Most of the uncertainty pertains to
the necessary assumption that empirical methods result
in uniformly labeled soil mineral N pools. Underestima-

1714 SOIL SCI. soc. AM. 1.. VOL 63. NOVEMBER—DECEMBER 1999

"N2 molecular fraction II

 

0 0.1 0.2 0.3 0.4

 

0.5 0.6 0.7 0.8 0.9 1

1sN atomic fraction .1

Fig.4. Visual'mtionofa methodtoc'ne-ventthe

ofpoolnon-nnifornn'ty. Alter-n

”Alibi.”

proble- nitnliadian'oa' s-led.
thechamberisnmheddM-dwiltedwith'N,AzisdisplaoedrelaﬁvetoApMme-entofm,p'veathehjeaoryofaaecnnlrny
pr'ndpalm—ptionis chm-acted

thatconvergesontheﬁlstatP‘ﬂThe

tion is expected to result from the failure of the assump-
tion, but the magnitude of the underestimation is difﬁ-
cult to predict.

Our work does not imply the existence of a theoretical
method 10 t heﬁeld. “Sta-
tistical expectation", as used in our argument has a
precise mathematical deﬁnition that is not equivalent
to expected error in ﬁeld measurements, unless ﬁeld
conditions closely match the constraints of the mathe-
matical model. The underestimation mlculated by our
equations for e will not likely be realized in the field
unless (i) the number Of pools is large, (ii) the enrich-
ments Of the pools are randomly distributed. and (iii)
flux is distributed evenly among the pools. We doubt
that any of these conditions is likely to be met in ﬁeld
settings. especially the third. Even if the conditions were
met. it would seem impossible to know this a priori.
Our experience with NO fluxes shows underestimation
to vary within experiments and especially among experi-
ments. When
pared to estimates made by gas chromatography. agree-
ment ranged from 6 to 117% (MS/GC, unpublished
data). In a systematic laboratory study of N20 fluxes,
Mulvaney (1988) found that differences between MS
and GC usually were small (less than 10%) and probably
resulted from analytical error. Mulvaney and Vanden
Heuvel (1988) found no appreciable difference between
MS and GC in the ﬁeld. unless plots were relabeled.
Can pool uniformity ever be assumed? We believe that
when added N (labeled) far exceeds preexisting soil
pool N. there is initially only one signiﬁcant pool, which
is practically uniform.

When uniformity of the soil N pool cannot be as-
sumed, it should be demonstrated (e g., Stevens et al.,
1997). However. the conceptual model given here (e. g...
Fig. 2 and 3) reminds us that it is not critical to know
that the soil pool' Is uniform; rather it is critical to know
the isotopic character of N gas derived from the soil
pool. Uniformity merely makes this easy to calculate

ctimatoc are ram

26

“personals-that

theevolvingp. (Soda k We tidy.)
(by invoking the equilibrium curve as one constraint).
The error that attends violation of the assumption of
pool uniformity could be avoided if there were altema-
tive means of assessing the isotopic character of the N
gas derived from the soil.

It maybeposs ible ' ‘ ‘ ' L ‘
character of soil-derived gas. Consider two successive
incubations of the same unit of soil. Suppose that, after
sampling the headspace at the end Of the ﬁrst incubation,
the enclosure is flushed. closed, and spiked with a small
amount Of ”N, The spike has the effect of displacing
A1. the base of the ray for the second incubation (Fig.
4). The rays for the ’ ‘ c. 5..
at a point identifying the isotopic character Ofw the total
evolved gas allowing accurate determinations of flux
for both intervals (equations are outlined in Note 4).
The principal assumption is that the isotopic character
of the evolved gas is constant: probably a more robust
assumption than pool uniformity. Another assumption
is that change in headspace enrichment due to other
processes (e. g., equilibration with soil pore space) is
negligible. The amount of gas needed for the spike de-
pends on the volume of the headspace and the sensitivity
of the mass spectrometer. For N,, such a spike will be
relatively expensive until ”N1 costs drop or sensitivity
improves such that smaller changes In ”N; abundance
become detectable. For N10, however, for which addi-
tions of labeled N such as “NO can readily be measured
in a normal atmosphere, this approach is already an
option.

Another approach for independently assessing the
isotopic character of soil-derived gas is similar to that
above. Two consecutive incubations are conducted, but
before the second incubation, the headspace is purged
of the gas of' Interest so that the ﬁnal sample will contain
only (mostly) soil-derived gas. This is, again, difﬁth
for N2 because of contamination problems, but is an
interesting possibility for N20.

We believe the heuristic model described here' is use-

BERGSMA ET AL: HEURISTIC MODEL or NITROGEN-GAS FLUX FROM NITROGEN-Is-LABELEO SOIL 1715

ful for clarifying principles. for designing experiments.
and for evaluating data related to N: or NO collected
over labeled soil. "N approaches to measuring N2 and
N20 fluxes are likely to become more common as associ-
ated materials and technology become more readily
available. Application of conceptual tools to explore the
limits and potential of isotopic methods is warranted.

Note 1

A Cartesian plot of 2"I: vs. ”a is graphically identical
to a ternary plot Of 2‘sx. 2"x. and 30x if the ternary plot is
bounded by an equilateral triangle. and if the abscissa is
expanded by a factor Of cos(30) " relative to the ordinate
(about 15%). Equivalence. but not identity. is preserved
even if both conditions are removed.

Consider an equilateral triangle Of unit height and
horizontal base. Specify three axes that bisect the three
vertices and intersect their respective Opposite sides at
right angles. Scale these from O to 1. base to vertex.
Beginning with the lower left vertex and proceeding
clockwise. assign the axis bisecting the vertex to repre-
sent ”it. ”x. or ”1:. respectively. Additionally. specify a
vertical axis and a horizontal axis (v and h) originating
at the lower left vertex and scaled identically to the
other three axes. Adopt the term “base“ to represent
a line normal to an axis. which passes through its origin.

For any given point. what is the relationship between
its Cartesian coordinates (h. v) and its ternary coordi-
nates (”x. ”x. 3".r)? It is clear by inspection that

v=2°x

because their bases are collinear. The distance Of any
point to the base of the h axis can be divided into
two portions falling inside and outside the ternary plot.
respectively. Trigonometric analysis shows that

h = (”x) cos(30)’l + (”1) tan(30).
Multiplying both sides by cos(30).

(h) cos(30) = (’° x) + (2" x) sin(30).
Since all 15N atoms occur either in the 29x or 30.: frac-

tions. and since only half of the atoms in the 29x fraction
are ”N, we can write

15a = (”x) + (”x) (0.5).
Noting that sin(30) evaluates to 0.5.
(h) cos(30) = l5a and h = ('50) cos(30)"

Therefore. a plot of (”x) vs. (”a) cos(30)" is identical
to a ternary plot of 23x, ”x, and 30x. The factor cos(30)“
merely scales the abscissa. and is largely irrelevant. The
plot ”x vs. 1’a also emulates a valid ternary plot. albeit
with a compressed base.

Note 2

To illustrate this. we calculated the isotopic character
of 1000 simulated gas mixtures by mass balance. Number
of pools was randomized on the interval [2.50]. Enrich-
ment as well as relative weights of pools were random-
ized on the interval [0.1]. When plotted. the isotopic

27

characters of the mixtures clustered around the coordi-
nates (1/2, 1/3) for (”a. 2"1). Mean coordinates for all
values were (0.497. 0.334). Mean coordinates for 100 000
simulated mixtures were (0.5004, 0.3336).

Note 3

In other words. h can approach arbitrarily close to 7.
and underestimation remains unchanged. When h = 7.
however, e is undeﬁned. The apparent suggestion is that
an almost perfectly uniform pool still leads to signiﬁcant
underestimation, when It is very close to y. The point
is moot. since flux estimates could hardly be made from
such a poorly labeled pool. Anyway, the reader is cau-
tioned that this result only arises under the assumptions
stated. The proof follows.

What is e for a mixture of equilibrium N, samples
whose enrichments are uniformly distributed on the in-
terval [7, h]?

DEW.
7:150
q=nxa=27-272
k=lsap

i= E("a) = ('y + h)/2

j=E(” x)=2i-2[i’+(h -y)2/12]

k = l-(m ' 2) I [(m - 2Y- 8(4 M7)]"’]/4
m = (i - (1)/(1' - 7)

e = (i - 7)/(k - 7)

Substitutions

j= 2 x [(y + h)l2] - 2[(y + h)2/4 + (h - y)2/12] =
(~4h2 — 472 - 4h‘y + 67 + 6h)/6
i-q=(-4h’-47’-4h7+67+6h)/6-(27-
272) = (-4h’ + 87’ — 4m - tn, + 6h)/6
m=(1"“(I)/[(7+h)f2*-7]=(-4h’+87z -4h7
- 67 + 6h)/l3(h - 7)]
m-2=[(-4h2+8y’—4hy—6ry+6h)—2(3h
- 37)]/[3(h - 7)] = (-4’!2 + 872 -4h7)/l3(h - 7)]
q 7=27 272-l(-4h’+87’- 4h7-67+
6h)/[3(h- 7)]17 = (- 27’ 272’! + 47h’)/I3(h- 7)]
(m - 2)2= l('-4h2 + 87’— 4h7)/[3(h - 7)]!2 = (1611‘
+647‘ -48h’7’+32h’7-64h7‘)/l9(h-7)’]
(m-2)’-8(q-M7)=(16h‘+647‘-48h’7’+32h’7
- 64h7’)/[9(h - 7)’] - 8(-27J - 27”! + 47W)/I3(h
- 7)] = (16h‘ + 161‘ + — 64th - My’hy
[9(h - 7)’] = (411’ + 472 - 8h7)’/l9(h - 7)’l
[(m - 2)2 - 8(4 - m7)]"2 = (4'!2 + 47’
- 8h7)/[3(h - 7)]
= l-(m - 2) I [(m - 2)2 - 8 (q - m7)ll/2]/4 =
l-l(-4h2 + 872 - 4h7)/l3(h - 7)]1 I [(4122 + 472
- 8h7)/[3(h - 7)]11/4

1716 SOIL SCI. soc AM 1.. VOL 63. NOVEMBER-DECEMBER 1999

k. = (4h2 - 81’ + 4127 + 4h2 + 47’ - 8h‘y)/[12(h - 7)]
= (219 - 7’ - h7)/[3(h - 7)] = (7 + 2W3
k,=(4h’-873+4h-y-4h2-4-f
+ 3h7)/[12(h - 7)] = 7
For k,:

e = (i - 7)/(k - 7) (undeﬁned)
For k.:

e = (i - 7)/(k - 7) = [(h + 7 - 27)/2]/l[(7 + 210/3]
‘ ‘Yl = [(h " 7W] x {3’0}! - 27)] = [3(h — 7)]/
[40! - 7)] = 3/4

Note 4

Using the symbology in Fig. 4, fractional contribution
of the soil pool to the ﬁnal mix is calculated as

dn = (lsaMn '— ”(Rd/(”0P - 150*")

where dis the fractional contribution. n is the incubation
number. “a is the enrichment Of the sample. A is the
initial chamber headspace. M is the ﬁnal mix. and P is
the soil-derived component. The enrichment of P can
be found by solving for the intersection Of the two rays
as follows:

(50050) + C1 = (Sz)("a) + C2
("0)(s. ' 52) = C2 ’ Cl
”0 = (C2 ' CI)/(Si ‘ 52)

where sis slope, C is intercept (by analogy to Eq. [2a])
and the subscripts reference the two incubations.
Slope is calculated as

3n = (293m "' ”rm/("am ‘ 150M)

where ”x is the mole fraction of singly-substituted mol-
ecules.
Intercept is calculated as

C = ”x... - 50’0".) = ”1A.. - 50’0“.)-

ACKNOWLEDGMENTS

We thank S.K. Hamilton and EA. Paul for encouragement
and direction during various stages of manuscript preparation.
We are grateful to Matt Emmons for technical insights regard-
ing N-gas analysis. We are indebted to four anonymous review-
ers whose comments greatly improved the manuscript. This
work is supported by grants from the USDA NR] and TECO
Programs. the NSF RTG and LTER Programs, and the Michi-
gan Agricultural Experiment Station.

REFERENCES

Arah. J.R.M. 1992. New formulae for mass spectrometric analysis of
nitrous oxide and dinitrogen emissions. Soil Sci. Soc. Am. J. 56:
795—80).

Arah. J.R.M. 1997. Apportioning nitrous oxide fluxes between nitriﬁ-

28

cation and denitriﬁcation using gas-phue mass spectrometry. Soil.
Biol. BiochenL 291295-1299.

Boast. CW.. R.L Mulvaney. and P. Baveye. 1988. Evaluation of
nitrogen-15 tracer techniques fordirectmeasmementof denitrifica-
tion in soil: 1. Theory. Soil Sci. Soc. Am. J. 52:1317—1322.

Bouwman. A.F. 1990. Exchange of greenhouse gases between terres-
trial ecosystems and the atmosphere. p. 61-127. In A.F. Bouwman
(ed.) Soils and the greenhouse effect. John Wiley & Sons. New
York.

Brooks. P.D.. DJ. Herman. GJ. Atkins. SJ. Proser. and A. Barrie.
1993. Rapid. isotopic analysis of selected soil gases at atmospheric
concentrations. p. 193-202. In LA. Harper et al. (ed) Agriailtural
ecosystem effects on trace gases and global climate change. ASA
Spec. Pub]. 55. ASA. Madison. WI.

Eichner. MJ. 1990. Nitrous oxide from fertilized soils: summary of
available data. J. Environ. Qual. 19:272-2m.

Focht. DD. 1985. Differences in nitrogen-15 enrichments of evolved
nitrous oxide and dinitrogen and the question of a uniform nitrate-
15 pool. Soil Sci. Soc. Am. J. 49786-787.

Hauck. R.D..and D.R. Bouldin. 1961. Distribution of isotopic '
in nitrogen gas during denitriﬁcation. Nature (London) 191:
871-872.

Hauck. R.D.. SW. Melsted. and RE. Yankwich. 1958. Use of N-iso—
rope distribution in nitrogen gas in the study of denitriﬁcation. Soil
Sci 86287-291.

Intergovernmental Panel on Climate Change. 1996. J.T. Houghton
et al. (ed) Climate change 1995: The Science of climate change.
Cambridge University Press. Cambridge.

Mosier. A.R.. W.D. Guenzi. and EE. Schweizcr. 1986. Soil losses of
dinitrogen and nitrous oxide from irrigated corps in northeastern
Colorado. Soil Sci. Soc. Am J. 50:344-348.

Mosier. AR. and L Klemedtsson. 1994. Measuring denitrification
in the ﬁeld. p. 1047—1066. In R.W. Weaver et al. (ed) Methods of
soil analysis. Pan 2: Microbiological and biochemical Properties.
SSSA Book Series. no. 5.x. SSSA. Madison. WI.

Mulvaney. RI. 1984. Determination of "N-labeled dinitrogen and
nitrous oxide with triple collector mas spectrometers Soil Sci.
Soc. Am. J. 48:690-692.

Mulvaney. R.L. 1988. Evaluation of nitrogen.15 tracer techniques
for direct measurement of denitriﬁcation in soil: Ill. laboratory
studies. Soil Sci. Soc Am. J. 52:1327—1332.

Mulvaney. R.L. and CW. Boast. 1986. Equations for determination
of nitrogen-15 labeled dinitrogen and nitrom oxide by mass spec-
trometry. Soil Sci. Soc. Ant J. 50360—363.

Mulvaney. R.L.. and LT. Kurt; 1982 A new method for determina-
tion Of "N-labeled nitrous oxide. Soil Sci. Soc. Am. J. 46:1178-1184.

Mulvaney. RJ.. and LT. Kurt; 1985. Reply to “Differences in nitro-
gen-15 enrichments of evolved nitrous oxide and dinitrogen and
the question of a uniform nitrate-15 pool”. Soil Sci. Soc. Am.
J. 49:787.

Mulvaney. KL. and R.M. Vanden Heuvel. I988. Evaltntion of nitro-
gen~15 tracer techniques for direct measurement of denitrifiation
in soil: IV. Field Studies. Soil Sd. Soc. Am. J. 52:1332-1337.

Siegel. R.S., RD. Hauck. and LT. Kurtz. 1982. Determination of "N,
and application to measurement of N: evolution during denitrifica-
tion. Soil Sci. Soc. Am. J. 46178-74.

Stevens. RJ.. RJ. Laughlin. GJ. Atkins. and SJ. Prosser. 1993. Auto-
mated deterrnination Of nitrogen-IS-labeled dinitrogen and nitrous
oxide by mass spectrometry. Soil Sci. Soc. Am. J. 57:981-988.

Stevens. RJ.. RJ. Laughlin. LC. Burns. and J .R.M. Arah. 1997. Men-
suring the contributions of nitriﬁcation and denitriﬁcation to the
flux of nitrous oxide from soil. Soil Biol Biochcm 29:139—151.

Vanden Heuvel. R.M.. R.L. Mulvaney. and KO. Hoeft. 1988. Evalua-
tion of nitrogen-15 tracer techniques for direct measurement of
denitriﬁcation in soil: 1]. Simulation studies. Soil Sci. Soc Am.
J. 52:1322-1326.

Weier. K.L.. J .W. Doran. J .F. Power. and D.T. Walters. 1993. Denitriﬁ-
cation and the dinitrogen/niacin oxide ratio as affected by soil
water. available carbon. and nitrate. Soil Sci. Soc. Am. J. 57:66-72.

Chapter 3

A NOVEL METHOD FOR DIRECT DETERMINATION OF NITROUS OXIDE AND
DINITROGEN FLUX FROM 15N-LABELED SOIL

Summary

Field measurements of denitriﬁcation gas products are useful for
studying both the contribution Of nitrous oxide to global climate change and the
loss Of fertilizer nitrogen as nitrous oxide or dinitrogen. Understanding the
relationship between N2 and N20 ﬂuxes may lead to process-based models
that improve prediction of N gas ﬂuxes. We report here an analytical method
that uses a single conceptual approach for independent analysis Of N2 and N20
for the same ﬁeld incubation. Sequential samples are removed from a
chamber cover placed over 15N-labeled soil. A small amount Of each sample is
analyzed for the isotopic composition of N2, and the remainder for N20. Each
aliquot passes through an ascarite trap to remove water and a LiOH trap to
remove 002; N20 is cryogenically focused in a liquid nitrogen trap. Aliquots are
then carried by He through chromatographic columns to further resolve
interfering masses. Finally, a triple-collector mass spectrometer detects peaks
for mlz 28, 29, and 30 (N2) or 44, 45, and 46 (N20). The peaks are integrated
and expressed as ratios relative to a laboratory standard. Flux is determined by
the change in isotopic character of chamber N2 or N20 with time. Unlike other
published methods for N2, 02 is separated chromatographically rather than

removed chemically. For N20, this is the ﬁrst report Of a method that infers ﬂux

29

directly from the change in isotopic character of headspace N20. Since the
method is equivalent for the two denitriﬁcation products, a subsequent
estimate of N20 mole fraction (N20 1 [N20 + N2] ) may be free of a well-known
systematic error that attends such methods when gases derive from non-
uniforrnly enriched substrate.

Introduction

The observed rate of increase in the concentration of N20 in the Earth's
atmosphere of 0.25% per year has raised concem over its contribution to
global warming (IPCC 1996) and to the destruction of stratospheric ozone
(Hahn and Crutzen 1982). A signiﬁcant proportion of the increase in N20 may
derive from terrestrial environments (Bouwman 1990), especially as a
consequence of agricultural activity (Eichner 1990, Mosier et al. 1998).
Denitrification and nitriﬁcation are responsible for most N20 ﬂux from soils
(Firestone and Davidson 1989) although their relative importance is still
unclear (e.g. Stevens et al. 1997).

Microbial denitrification in soils (Paul and Clark 1996, Robertson 1999)
results in variable proportions of its end-products, N20 and N2. The
relationship between denitriﬁer N20 production and total denitriﬁcation is
conveniently deﬁned as the N20 mole fraction: N20 I [N20 + N2]. The N20 mole
fraction is highly variable spatially and temporally (Letey et al. 1980, Hutchinson
and Davidson 1993, Weier et al. 1993). Single estimates of N20 mole fraction
cannot be used reliably to relate denitriﬁcation and N20 ﬂux (Vinther 1984,

Aulakh et al. 1992, Weier et al. 1993) because of the variability in mole fraction.

30

Improved methods of measuring N2 and N20 emissions from soil can foster
the understanding of global N20 emissions, loss of fertilizer N from the plant-
soil system during denitriﬁcation, and the relative contributions of nitriﬁcation
and denitriﬁcation to soil N20 ﬂux.

Direct methods for measuring fluxes of both gases are few. Chamber
techniques prevail for N20 (Mosier 1989) because its concentration is readily
determined by gas chromatography using an electron-capture detector.
Dinitrogen can also be quantiﬁed by gas chromatography, but owing to its
abundance in the atmosphere its concentration is usually not detectably altered
during chamber incubations. The acetylene inhibition technique (AIT) is often
employed to measure total denitriﬁcation gas flux (e.g. Mosier and
Klemedtsson 1994) since acetylene blocks reduction of N20 to N2 (Yoshinari
and Knowles 1976). Dinitrogen ﬂux can be determined by difference, if a
control chamber or soil core is used to measure N20 alone; however, the
potential variability between chambers introduces uncertainty, which must be
accommodated by adequate replication. Also, AlT may be ineffective for heavy
textured soils (Arah et al. 1993). Furthermore, acetylene blocks nitriﬁcation at
levels lower than those that inhibit N20 reductase, which can cause
denitriﬁcation to be underestimated because of substrate depletion (Bollmann
and Conrad 1997). Finally, some microorganisms metabolize acetylene (Topp
and Germon 1986).

An alternative method for determining N2 ﬂux is to label soil with 15N03' or

”NH; and analyze a time series of chamber gas samples by isotope ratio

31

mass spectrometry (IRMS) (Hauck and Bouldin 1961, Siegel et al. 1982). This
method, known as the 15N isotope dilution technique, is becoming more
common (Mulvaney and Kurtz 1984, Mosier et al. 1986, Stevens et al. 1993).
One modiﬁcation involves reducing N20 to N2 chemically before analysis to
measure total denitriﬁcation (Mosier et al. 1990, Arah et al. 1993, Mosier and
Klemedtsson 1994). If total denitriﬁcation is measured, the N20 mole fraction
can be determined if N20 ﬂux is independently assessed (e.g. Arah et al. 1993).
Dinitrogen ﬂux has also been determined by radioactive (‘3N) isotopic labeling
(Firestone et al. 1979, Speir et al. 1995) but the radioactive tracer is short-lived
and limited in availability.

The procedure described here estimates ﬂuxes of N20 and N; from soil
independently by detecting changes in isotopic composition of headspace N20
and N2. Data from a ﬁeld experiment are provided as an illustration. This is the
ﬁrst demonstration of N20 and N2 flux concurrently measured by equivalent
methods. Similarity of method and sampling of the same headspace may
reduce error in the estimate of the N20 mole ratio.

Materials and Methods
Overview

The method may be summarized as follows. ‘5N03' is added in
aqueous solution to soil. A headspace is conﬁned over the soil surface and
N20 and N2 are allowed to accumulate (eg. by denitriﬁcation). The headspace

is sampled using evacuated glass vessels fitted with stopcocks. Samples are

32

puriﬁed in the laboratory, and then analyzed by isotope ratio mass
spectrometry.

'lsotopic character is defined as the relative abundances of all isotopic
masses of a gas (Bergsma et al. 1999). The shift in isotopic character of a
headspace gas (either N; or N20) indicates the relative contributions (to the
ﬁnal mixture) of soil-derived gas and atmospheric gas initially present.
Changes in isotopic character also indicate the enrichment of the soil N pool
undergoing denitriﬁcation. The rate of emission can be calculated from the
relative contribution of the soil source as a function of elapsed time, chamber
area, headspace volume, and initial or ﬁnal concentration. The initial
concentration of N20 or N2 may be assumed, but for N20 it is more accurate
(and not very difﬁcult) to obtain a measurement.

Vessels and Sampling

Pyrex vessels are used to collect headspace samples. The 500-mL
volume of the vessels provides sufficient N20 for analysis at ambient
concentrations in air. Analysis of N2 is performed on a subsample from each
vessel, prior to N20 analysis. The vessels are oblong, with stopcocks at each
end. The stopcocks, constructed of glass barrels and pistons, are sealed with
VITON O-rings and terminate in 1/ " (6.4 mm) 0.0. glass tubing. Vessels are
pre-evacuated to a pressure of less than 1.3 Pa. In the ﬁeld, the vessels are
ﬁtted to a chamber cover using latex tubing. Samples are collected by rapidly

opening the chamber-side stopcock for ~10 s.

33

Analysis

For isotopic characterization of N2 and N20, samples are processed
using a preparation system interfaced to a mass spectrometer (Figure 3.1).
Sample vessels are attached to the terminus of the system using 1/ " ID
CAJON Ultratorr unions. The system is then evacuated. For analysis of N2, the
sample trap is isolated from vacuum; the vessel stopcock is opened for 10 s
and sample diffuses into the sample trap (0.5 m x 1116" large internal diameter
nickel tubing, coiled). The remaining sample (the majority) is reserved in the
vessel for later N20 analysis. For both gases, the sample passes through a 10
cm column of ascarite (10 mm ID.) for removal of water, then to a 20 cm
column (4 mm ID.) of anhydrous lithium hydroxide for removal of C02, which
has the same mass spectrum as N20.

For N20 analysis, the sample trap remains open to vacuum and the
complete contents of the sample vessel pass through the chemical scrubbers
at a regulated rate. In contrast to analysis of N2, however, the sample trap is
chilled with liquid nitrogen to retain N20. The evacuation of the vessel is
regulated by an electronic mass ﬂow controller (25 mL min“) to maintain the
efﬁciency of the sample trap. After trapping of either N2 or N20, the sample trap
is isolated by pneumatically operated valves (and then thawed, in the case of
N20). A stream of He (50 psi head pressure) then carries the contents of the
trap onto the column of a gas chromatograph. For analysis of N20, a J.W.
Scientiﬁc GS-Q column is used to assure separation from C02. A molecular

sieve column (Alltech, 8 m by 1/ " O.D., 5A) is used to separate N2 from 002, Ar,

34

and 02 (oven temperature, 50° C). The column terminates at the inlet of the
isotope-ratio mass spectrometer. Prior to arrival of a sample peak, the mass
spectrometer analyzes a pulse of reference gas (Note 1). The ratios 2/1 and
3/1 are calculated for both the sample and reference by routines which
integrate the signals from beams 1,2, and 3 (masses 44, 45, and 46 for N20 or
28,29, and 30 for N2). For N2 analysis, a pressure regulator adds make-up
helium (30 psi) into the ﬂow path just before the inlet of the mass spectrometer
(Figure 3.1) to dilute the N; for optimal peak size.

Because of the low abundance of N20 and high abundance of N2 in air,
typical ﬂuxes from an isotopically enriched source (eg. soil N03) to a chamber
headspace cause a marked increase in the 15N enrichment of N20, but only
mildly perturb the level of enrichment in N2. Therefore, the potential range of
isotope ratios is greater for N20 than for N2. The Micromass Prism mass
spectrometer used in this study has a second head ampliﬁer that is readily
interchanged with the ﬁrst. Initially these were both conﬁgured such that the
two minor ion beams were ampliﬁed 100 times more than the major beam
(resistor values 5 - 108 Q, 5 - 1010 Q, and 5 - 1010 a). This conﬁguration
anticipates that the minor beams will be much smaller than the major, which is
usually the case for natural abundance measurements and is still the case for
our analysis of N2 in our chamber experiments. We modiﬁed our second head
ampliﬁer with the result that all three resistors were of equal but intermediate

sensitivity (1 - 109 (2). This conﬁguration makes no assumptions about the

relative strengths of the major and minor beams and is appropriate for

35

enriched gases (>596). Normally all samples in a set are analyzed for N2
before reconﬁguring the mass spectrometer for N20.
Data Processing

The equations of Arah (1992) were modiﬁed by Bergsma et al. (1999) to
calculate ﬂuxes and estimate enrichments of the gas source. Application of
these equations to N20 is completely analogous to their use for N2. However,
because of naturally occurring isotopes of oxygen, the molecular fractions
“N20 and 46N20 do not strictly correspond to the molecular fractions
[“N‘5N0+‘5N14NO] and 15N‘sNO. We have derived equations that express
29(N.,.)O/"‘(N2)O and 3°(N2)0128(N2)O as functions of “NZO/“Nzo, 46Nzo/“Nzo,
"OI“O and 18OI"’O; thus the equations for N; can be used directly for N20 (Note
2).

Since the equations estimate d (the fraction of mixed gas derived from
the soil source), estimation of absolute flux requires some estimate of
absolute abundance. If a represents gas from the atmosphere (pre-existing

headspace gas) and p represents gas derived from the soil mineral pool, then

d=pl[a+p]. [1]
Therefore

p = d * [a + p] ; alternatively [2]
p = da / [1-d]. [3]

Since concentration and therefore actual volume of headspace N2 hardly
changes during a typical incubation,

[a + p] z a and therefore [5]

36

p =- da. [6]

For N20 ﬂux, however, Eq. [3] must be used rather than Eq. [6]; a is the
average abundance of N20 in the atmosphere (~3.1 - 10'9 L - L") multiplied by
chamber volume. Alternatively, concentration can be measured by gas
chromatography at the beginning and end of an incubation, so exact values of a
(beginning; Eq. [3]) and [a + p] (end; Eq. [2]) are known. In the ﬁeld experiment
described later, similar results were obtained whether Eq. [3] was used with an
estimated a, Eq. [3] was used with a measured a, or Eq. [2] was used with a
measured [a + p].

Veriﬁcation

Tests were conducted to assess the performance of the mass
spectrometer and gas puriﬁcation system. To determine whether isotopic
analysis is inﬂuenced by the size of the sample, replicate samples of N20
reference gas were analyzed using the traditional head ampliﬁer at four
different intensities for the major beam, covering the range of valid sample
sizes for this system (Table 3.1 ).

Our mass spectrometer is designed for analysis of samples having
isotope ratios near natural abundance. To test for a memory effect (i.e.,
whether analysis of highly enriched samples potentially biased the analysis of
subsequent samples by contaminating some portion of the system with
residual 15N) we analyzed a sample of N20 at natural abundance (45l44 only)
immediately following analysis of a highly enriched sample. Enriched samples

were prepared by mixing natural abundance N20 with 45N20, as described

37

elsewhere in this study. Samples were injected by gas tight syringe. Measured
enrichments are reported in 6 (per mil) notation (Table 3.2) to facilitate
interpretation:

5=[(RSAM/RSTD)'1]1OOO

where R is the ratio of mlz 45 to 44, RSAM is the sample ratio, and Rsm is
the ratio of the standard. Since the laboratory standard is identical to the
reference gas, a measurement of 0 %o (0 per mil) was expected for the
standard gas, and the subscripts refer to the sample and the standard.

We tested the response of the mass spectrometer for linearity over a
moderate range of enrichments using the traditional head amplifier. Natural
abundance N20 and puriﬁed 45N20 (>98% 15NMNO) were mixed on a vacuum
line. Natural abundance N20 was measured with a calibrated glass bulb (~20
mL) and enriched N20 was measured with a gas-tight syringe (0-10 uL).
Gases were frozen into a 1L ﬂask submerged in liquid nitrogen. After thawing,
aliquots of mixture could be removed from the ﬂask and analyzed, or the
mixture could be refrozen for further dilution or enrichment. Aliquots (~03 mL)
were analyzed using the preparation system described above. Results are
reported as a calibration line (Figure 3.2).

The response of the mass spectrometer was tested for linearity over a
wide range of enrichments using the modiﬁed head ampliﬁer. Natural
abundance N20, puriﬁed 45N20, and/or 46N20 were mixed in a ﬂask in ratios of
small whole numbers. In order to prevent previously observed fractionation

effects at low pressure, ﬂask pressure was increased to ~1 - 105 Pa (1 atm.) by

38

adding He (99.999%). Replicate aliquots were removed from the ﬂask for
analysis, with highly repeatable analytical results (Table 3.3).

The N20 analysis system was tested for long-term stability, for an effect
of the presence of a LIOH column on measured isotope ratios, and for an effect
of water in the LIOH column. A mixture of natural abundance N20, 45N20, and
48N20 was prepared in a ratio of 1:1 :2 and analyzed. Three weeks later, more
aliquots were analyzed using each of the following: no LiOH column, a hydrous
LIOH column, and a column of LiOH dried by heating on a vacuum line
overnight (Table 3.4). The between-date comparison is also a test of the
stability of laboratory mixtures.

Rapid movement of gas through narrow apertures can cause
fractionation. We tested whether significant fractionation occurs during ﬁlling of
the Pyrex sampling vessels described previously. Three vessels were left
open in the laboratory for several days, and three were evacuated and ﬁlled
rapidly at the time and location that the other three were closed. N20 in the
vessels was analyzed as described, using the modiﬁed head ampliﬁer (Table
3.5).

Laboratory Denitriﬁcation

The equations used for data processing allow calculation of the mean
isotopic enrichment of the soil mineral N pool from which evolved N2 or N20 is
derived. We conducted a laboratory experiment to verify the accuracy of
calculated enrichments. Fresh soil was collected in July 1998 from a wheat

plot at the Kellogg Biological Station, Ml, that had not been fertilized at planting.

39

Water-extractable NO; in this soil was ~2.8 pg - g‘1 dry soil. Ten 9 of fresh
sieved soil (8% moisture) were added to each of fourteen 600 mL Erlenmeyer
ﬂasks ﬁtted with septa. Each ﬂask also received ~1 g steel wool activated with
detergent solution (as a sink for trace levels of 02 - Parker, 1955; cited in
Kaspar and Tiedje, 1994) and 10 mL of ~1.0 mM sodium succinate. Flasks
were ﬁtted with evacuated sampling vessels using one-holed stoppers, then
ﬂushed with high purity nitrogen (via the septa, using a source needle and vent
needle) and monitored for N20 production. Quantitative consumption of native
soil NO; was suggested by cessation of N20 accumulation and conﬁrmed by
selective destructive sampling for soil N031 Stock solutions of 99.93 atom
percent 15N-KN03 and natural abundance KN03 were prepared and mixed to
give secondary solutions of ~0, 10, 20, 40, or 100% 1SN target enrichment.
Each ﬂask was ﬂushed with N2, and then received 1 mL of a secondary
solution, to deliver ~5 pg N as N031 When the N20 concentration in the ﬂask
headspace reached 1-2 ppm, samples for isotopic analysis were collected by
opening the vessel stopcocks. Samples were analyzed within two days, plotted
as 1"’x vs. 15a (Figure 3.3) and tabulated (Table 3.6). The process was repeated
for analysis of N2 production, using 20 g fresh soil per ﬂask (1.4 pg N03'-N - g
dry soil"), 20 mL H20, 1 mL of 0.1 M sodium succinate. and 1 mL of 0.1 M
secondary solution. Flasks were ﬂushed thoroughly with high purity N2. Four
blank vessels were ﬂushed with N2 as a reference. Evolution and subsequent
disappearance of N20 in the headspace suggested active denitriﬁcation.

Anaerobic conditions were confirmed by monitoring headspace 02 by gas

40

chromatography. After several days, headspaces were sampled by opening
vessel stopcocks. Destructive sampling of one vessel showed ~2 ug N as
N03' remaining (cs. ~1.4 mg added). N2 in sample vessels was analyzed for
isotopic abundances. Average enrichments were calculated by the equations
of Bergsma et al. (1999), using the average isotopic character of the reference
ﬂasks to represent initial headspace (Table 3.6).
Field Demonstration

In April 1998 a ﬁeld of winter wheat at Kellogg Biological Station, MI was
fertilized with NH4NO3 at a rate of 30 kg N - ha". Fertilizer was excluded from six
microplots (0.25 m2) nested within treatment plots (32 m2), three of which were
clipped. The microplots received 99% ‘5N-KN03 at a rate of 30 kg N - ha".
Aluminum frames (0.0846 m2) were installed in each microplot as bases for
gas sampling chambers. A clipped microplot was selected for intensive study.

Rain fell sporadically over the two week period following fertilization.
Nine one-hour incubations were distributed around these rain events. A lid (30
cm x 30 cm x 14 cm) was placed over the selected frame (sealed with a moat of
water). At the beginning and end of each incubation, gas samples were
collected for analysis by gas chromatography (GC), infrared gas analysis
(IRGA), and mass spectrometry (IRMS). Samples for CC and IRGA were
collected by syringe to 3 mL Vacutainers (4 mL overpressure). For MS
analysis, pre-evacuated glass vessels were connected to the chamber lid

using latex hose and opened at appropriate intervals.

41

To minimize the pressure artifact at the soil surface associated with
opening the evacuated vessels, "barostatic" chamber lids were used. Each
barostat consists of a resealable polyethylene bag (44.4 um thick, 0.94 L
capacity) and a short length of threaded pipe (ID. 7 mm) which attaches one
wall of the bag to an internal wall of the lid (using a washer and 0-ring) and
vents the bag to the exterior of the lid. The bag can be opened for assembly
and resealed for deployment. The pipe can be stoppered, except during
sample collection, to guard against leakage due to bag failure or diffusion.
Normally two barostats are used per lid, and are pressure-tested prior to each
incubation. For the experiment described above, only one unstoppered bag per
lid was used.

Results
Veriﬁcation

The data in Table 3.1 indicate that the measured isotope ratios for N20
reference gas do not vary appreciably with sample size on our mass
spectrometer although, as expected, larger samples provide better precision.
\M'ten equipped with the traditional head amplifier, the mass spectrometer
demonstrates linearity over a large range of moderate enrichments (Figure 3.2)
with only a slight bias in favor of the heavy isotope (slope ~1.05). Apparently
our system slightly overestimates the enrichment of a natural abundance
sample immediately following analysis of a highly enriched sample (Table 3.2).

In our judgment, the overestimation is negligible.

42

Agreement of measured and calculated values was usually within 1% for
high levels of enrichment using the modiﬁed head amplifier (Table 3.3). We
could not ﬁnd evidence that the presence of LIOH or the presence of water in
the LiOH signiﬁcantly affects measured isotope values of N20 (Table 3.4; P > F
for the effect of LIOH condition on 45N20 and 46N20: 0.8306 and 0.8989,
respectively). Furthermore, measurements are stable across periods as long
as three weeks (P > F for the effect of date on 45N20 and 46N20: 0.4102 and
0.5362, respectively). We found no evidence that sample gas is signiﬁcantly
fractionated upon collection using evacuated vessels (Table 3.5). Repeatability
was better when gas was collected quickly by opening pre-evacuated vessels
(SE: 5.5- 10‘“, 2.4- 1043 for ion beam ratios 2/1 and 3/1, respectively) than upon
collection over long time intervals by diffusion of ambient air (1.4 X 10" and 1.9
x 1043 for 2/1 and 311, respectively).

Laboratory Denitriﬁcation

The measured and predicted enrichments for N20 and N2 produced from
labeled soil in ﬂasks agreed well (Table 3.6). The predictions are slightly
different from the 'target' enrichments because of the assumption of 0.3663%
15N in natural abundance KNOa (i.e., equal to the 15N abundance in atmospheric
N2) and a measured 0.07% 14N in stock 15N-KNO3. Despite evidence of
statistically significant differences (Student's ttest), absolute differences are
small: in all but one case the means of measured enrichments are within 1%

(0.01) of the predicted value. These results indicate the lower limit of the

43

accuracy of our method and eliminate systematic error as a possible
explanation for unusual ﬁeld results.

In the case of N20, the laboratory denitriﬁcation experiment provides a
test of an important assumption: namely, that microbial denitriﬁcation of a
uniform soil N pool results in an equilibrium mixture of masses (that is, N
atoms with masses 14 and 15 are paired in a statistically random fashion,
such that the abundances of singly-, doubly- or un-labeled molecules in the
product pool can be predicted from the enrichment of the substrate; see Note
3). Since no headspace N20 was initially present, sampled N20 was entirely
from a soil source with presumably uniform enrichment. Under these
circumstances, the sampled gas should be in equilibrium. Figure 3.3 shows
nine samples (three replicates at three enrichments) plotted with the
equilibrium curve (see Bergsma et al. 1999). Since the data points lie very near
the curve, the equilibrium assumption for microbial denitriﬁcation appears to be
supported. Previous studies have tested the equilibrium character of N20 by
comparing 15N content calculated from 45R and 46R (Stevens et al., 1997; 1998a;
1998b). Here, information from 45R vs. 46R is combined to calculate isotopic
character, a two-dimensional quantity, which can be compared to the
equilibrium curve, a two-dimensional reference.

Field Demonstration

The labeled soil experiment (April 1998) is a field application of the

analytical methods described above. Nitrous oxide and dinitrogen fluxes

showed similar temporal trends; however, during the ﬁrst two days of the

44

experiment N20 and N2 fluxes generally increased and decreased, respectively
(Figure 3.4). The associated rise in N20 mole fraction (~20/[N20 + N2], Figure
4.5) is counter-intuitive: accumulation of soil water due to precipitation should
decrease soil redox status and favor production of N2. Enrichment of the soil
pool undergoing denitriﬁcation was estimated non-destructively by analysis of
the shift in isotopic character of headspace gases during each incubation
(Hauck and Bouldin 1961, Bergsma et al. 1999). The apparent enrichment,
based on N20 isotope data, dropped from 0.82 to 0.72 during the 4 day period,
suggesting that significant nitriﬁcation was occurring. Soil pool enrichment
estimated from N2 data was considerably more variable, perhaps reflecting
lower sensitivity for N2 (see precision and detection limits, below). Agreement
of mass spectrometric and gas chromatographic flux estimates for N20 was
good (Figure 3.4), with 5 values of the MS/GC ratio between 94% and 107%,
and three values between 66 and 77%. Differences could be due to analytical
error, to non-uniform enrichment of the soil mineral N undergoing denitriﬁcation
(cf. Mulvaney et al. 1988), or to other sources of N20 production (e.g. Robertson
and Tiedje 1987, Stevens et al., 1998b).
Precision and Detection Limits

There is not as yet a universally recognized method for calculating
detection limits when denitrification fluxes are quantiﬁed from shifts in isotopic
character. However, means and standard deviations of ion ratios 2/1 and 3/1
(the primary output of isotope ratio mass spectrometers) are easily compared

(Table 3.7). Precision for 211 (N2) in this study is comparable to that in previous

45

reports, and for 3/1 is less. Precision for N20 and N2 in this study are similar.
Mean 3/1 ratios for N2 are substantially higher than those reported by others.
This is probably an artifact of high mass 30 background due to the formation of
NO in the spectrometer source (Stevens et al 1993). Post hoc correction of
data represented in Figure 3.4 using a representative calibration curve lowered
the mean N2 ﬂux by ~4%.

Detection limit has been deﬁned as three times the standard deviation of
the blank (Miller and Miller, 1988, cited in Stevens et al. 1993). Interpreting this
to mean that ratios 2/1 and 3/1 in end-of—incubation samples must exceed
initial ratios by three standard deviations (as tabulated), we calculated
minimum detectable flux independently for 2/1 and for 3/1, and then selected
the larger of the two. Detection limit is a function Of analytical precision,
chamber volume, chamber area, enrichment of the soil pool, and duration of
the incubation. For the ﬁeld demonstration above, headspace was ~14 L, area
was 0.0846 m2, enrichment averaged 0.77 15N, and duration was ~1 h;
consequently our estimated detection limit for N2 is 216 g - ha'1 - d". This is
larger than reported estimates of 5 g - ha'1 - cl'1 (Siegel et al. 1982) and 12 9 ~
ha'1 - of1 (Stevens et al. 1993), but not directly comparable because sampling
configurations were different. For N20, our detection limit was 2.72 - 10“ 9 N20-
N - ha'1 - d", which is equivalent to a minimum detectable change in
headspace concentration of 6 - 10'12 L - L". Stevens et al. (1993) report a
minimum detectable change in headspace concentration of 2.1 - 10'6 L - L'1 for

N20. For N20 analysis by gas chromatography, assuming a nominal CV of 1%

46

(perhaps optimistic), minimum detectable concentration change is ~ 1 - 10‘8 L-
L".
Discussion

Our results provide evidence that ﬂux of N20 and N2 from 15N-labeled soil
can be reliably measured by isotopic analysis of chamber headspace gases
using mass spectrometry to directly analyze N20 and N2 after chemical,
cryogenic, and chromatographic puriﬁcation. We veriﬁed that our analytical
technique gives reasonably accurate, precise, and unbiased results. The ﬁeld
data show that our analytical technique can be an integral part of a complete
experimental system. Although similar methods exist for N2, ﬂux of N20 has
never been analyzed in this way before. Usually N20 is reduced to N2 and
analyzed in a mixture with sample N2 (Mosier et al. 1990) or laboratory standard
N2 (Mulvaney and Kurtz 1982). Our method for N20 gives lower detection limits
and coordinates well with the 15N dilution technique for N2 because it involves
the same assumptions regarding homogeneity of the soil mineral pool.

An isotopic method for measuring total N; flux from soil has been
available for three decades (Hauck and Bouldin 1961) but it has only recently
been exploited (e.g. Siegel et al. 1982, Mulvaney and Kurtz 1984, Stevens et al.
1993). Our technique for estimating N2 ﬂux differs in some details from those
published. The conventional approach for purifying N2 involves removing
condensibles (e.g. C02) in a cold trap and removing oxygen with chemical
traps or hot copper (e.g. Siegel et al. 1982, Mosier and Klemedtsson 1994).

Boyd et al. (1994) claim superior convenience and efﬁciency for a mixture of

47

CaO granules and Cu for purifying nanomole quantities of N2 (mainly removing
C02 and 02, respectively). We remove water and the majority of C02 using
chemical and cold traps, and separate 0;, Ar, and trace 002 and CO from N2 on
a molecular sieve column during continuous flow mass spectrometry. Our
puriﬁcation system is efﬁcient and inexpensive, and requires very little
maintenance. The conceptual approach to the ﬂux calculation is in principal the
same as originally proposed (Hauck and Bouldin 1961) and uses the
equations and notation of Arah (1992) as modified by Bergsma et al. (1999).

In early studies of N2 ﬂux, 15N-IaoeIed fertilizer was added to soil and
evolved gas was evaluated for mass ratio 29/28 (e.g. Rolston et al. 1978,
1982). One limitation of this approach is that only the N2 ﬂux derived from
fertilizer is determined, and not any ﬂux derived from native soil mineral N (see
Mosier and Klemedtsson 1994). Another way of describing this limitation is
that the isotope ratio 29128 does not completely characterize N2 when the
sample is a mixture of atmospheric N2 and soil-derived N2 of different
enrichment; the ratio 30128 must also be measured (Hauck et al. 1958) since
such mixtures are not in isotopic equilibrium. Hauck and Bouldin (1961)
showed that measurement of the ratio 30I28 allows calculation of the average
enrichment of the N pool experiencing denitriﬁcation, which in turn allows a
calculation of total N2 ﬂux (whether from native soil N, label, or both). The
principle assumption of this approach is that ﬂux derives from a single,

uniformly labeled pool.

48

The N20 molecule, like the N2 molecule, can be singly or doubly
substituted with respect to 15N (giving masses 44, 45, and 46 instead of 28, 29,
and 30). Therefore, it is possible, as we have shown, to estimate ﬂux of N20 by
measuring shift in isotopic character (as for N2). This has not been performed
until now, probably because isotope ratio mass spectrometers, although more
precise than gas chromatographs, may not have been sufﬁciently sensitive. In
the past, as much as 0.1 mg N was needed for analysis, as compared to our
0.2 pg; see Mulvaney and Kurtz, 1982; 1985).

When measuring soil denitriﬁcation by chamber methods, it is usually of
interest to determine both N2 ﬂux and N20 ﬂux. Five methods have been
previously reported for determining N20 flux from soil otherwise labeled for N2
determination. Some researchers measure N2 ﬂux using the shift in isotopic
character ("Hauck technique") and N20 ﬂux by gas chromatography (Mosier et
al. 1986). Others measure N20 by GC and [ N20 + N2 ] by the Hauck technique,
reducing N20 over hot copper and thereby letting it mix with sample N2 before
analysis (Mosier et al. 1990, Arah et al. 1993). Third, some researchers
calibrate their isotope ratio mass spectrometer so that the concentration of N20
can be derived from the sum of ion currents above baseline for masses 44, 45,
and 46 (Stevens et al. 1993). A fourth approach is to trap N20 from the sample,
mix it with a known quantity of standard N2, and then reduce the N20 to N2
(Mulvaney and Kurtz 1982, Mulvaney and Kurtz 1984, Mulvaney and Vanden

Heuvel 1988). Finally, a linear mixing equation can estimate N20 ﬂux if the

49

enrichments of the soil mineral N, the label, and the evolved N20 have been
independently determined (Brooks et al. 1993).

Our method for quantifying ﬂux of N20 has advantages relative to the ﬁve
methods outlined above. The ﬁrst three methods of ﬂux estimation depend on
a change in headspace concentration, a measure of net flux, whereas isotope
methods generally measure gross fluxes (see discussion in Hart et al. 1994).
Net positive ﬂux of N20 is only the same as total ﬂux if signiﬁcant quantities of
N20 are not consumed during incubation; consumption of N20 has not been
widely tested, and may be especially important under soil covers, within which
N20 concentration (and therefore likelihood of consumption) is increasing. The
fourth and ﬁfth methods are isotopic approaches, but assume negligible
background N20 in the chamber headspace. The atmospheric background is
indeed small (~31 - 10'9 L - L“) but negligible only if concentration due to ﬂux is
several orders of magnitude greater. Clearly a method that does not assume
negligible background is potentially more sensitive. Our method requires
larger samples (500 mL) and longer analysis times (40 min/sample) than
those reported for automated methods (Brooks et al. 1993, Stevens et al. 1993)
but sensitivity is greatly enhanced, no destructive sampling of soil is required
(in contrast to the ﬁfth method above), and gross ﬂux rather than net ﬂux is
estimated.

In addition to the general advantages above, our method for
determination of N2 and N20 carries speciﬁc advantages for determination of

the N20 mole ratio during denitriﬁcation. (1) Because both N2 and N20 are

50

measured by shifts in isotope ratios rather than changing concentrations, both
ﬂux estimates represent gross rather than not ﬂow from the soil surface,
resulting in an internally-consistent ratio. (2) When the soil mineral pool is not
uniformly enriched, ﬂux of N2 is underestimated (see Boast et al. 1988, Vanden
Heuvel ef al. 1988, Arah 1992, Bergsma et al. 1999). To the extent that the
same mineral N pool is the source for both N20 and N2, proportional
underestimation of both will be similar when equivalent methods are used.
Therefore the calculated ratio of the two ﬂuxes will be relatively independent of
this source of error. (3) Unlike the acetylene inhibition technique, the method
described here and the others outlined above allow ﬂuxes of both N2 and N20
(and therefore the N20 mole ratio) to be determined from a single experimental
unit, thus reducing statistical uncertainty due to natural variability among control
and experimental units. The statistical need for replication is reduced.

Mile any given method for determining N20 ﬂux may have its
advantages, important gains are often made by coordinating multiple methods.
Mulvaney (1988) measured N20 by GC and by MS to test the assumption that
N20 was derived from a uniformly labeled pool of soil N. Arah et al. (1993)
used both GC and MS methods to test the suitability of acetylene inhibition for
measuring denitriﬁcation in heavy-textured soils. Stevens et al. (1997)
employed concentration and isotope distribution data from mass spectrometry
to examine relative contributions of nitriﬁcation and denitriﬁcation to N20 ﬂux.
As noted above, GC values that are lower than MS values may constitute

evidence of concurrent production and consumption of N20 by soil. As the

51

precision and convenience of these methods improve, so does the potential for

characterizing fundamental controls on denitriﬁcation dynamics. Such

improvements should lead to a more comprehensive perspective on regional

and global N budgets, in addition to local insight regarding soil N cycling.
Notes

Note 1. For N20 analysis, a well-characterized laboratory standard is
used. For N2 analysis, ~150 pL of 3°N2 are mixed with ~20 mL of laboratory
standard to improve the stability of 30l28, since natural abundance N2 has very
little mass 30. The absolute abundances of m/z 28, 29, and 30 are measured
manually: voltage shifts are used to sequentially place each mass in the same
collector for three replicate cycles. After correction for background readings,
the ratios of the means are calculated and used later to interpret the analytical
data for the samples analyzed with this reference.

Note 2. Let x, y, and 2 represent the fractional abundances of 2E’(N2)O,
”(N2)O, and 30(N2)O. Let r, s, and t represent the fractional abundances of
“N20, 45N20, and 46N20. Let c, d, and e represent the fractional abundances of
160,170, and 180. That the mass spectrometer measures slr and tlr, although
ylx and zlx are of interest. Now,

r=xc

s = yc + xd and

t = 20 + yd + xe.

Then

s/r = [yc + xd] /[xc] = ylx + d/c and

52

t/r = [20 + yd + xe]/[xc] = zlx + [yd]l[xc] + elc.

Thus,

ylx = slr - d/c and

zlx=tlr-y/x*dlc-elc.

In terms of simple mass ratios,

(”mac/”(Nam = [‘5N20/“N201 - I1 701‘ 601 and

I°°(N2)0/28(N2)Ol = Dime/“N201 - [29(N2)0/28(N2)OI * ["01‘601 - [mo/”0]-

ln shorthand,

29R = 45R - ”R and

NR = 46R _29R17R _ 18R
Literature values for ['7OI'GO] and [180/160] are used.

Note 3. Use of the term “equilibrium” in discussions of the 15N isotope
dilution technique is related to its classical use in discussions of chemical
reactions. The reaction

”N2 +3°N2 <-> 2I”N21
proceeds spontaneously in both directions, but at a negligible rate because of
high activation energies. Thus, the different molecular-mass fractions of an N2
sample do not normally equilibrate, except perhaps at geologic time scales.
The reaction rates increase, of course, if the sample is heated (> 1000 ° C) or
in the presence of some other form of energy (e.g. microwave). At chemical
equilibrium, the relative concentrations of products and reactants no longer
change. The relative proportions of the three mass fractions is then

approximately that which is predicted from the composite 15N abundance of the

53

whole system (neglecting fractionation effects). Any sample meeting this
criterion may be said to be in equilibrium, even if no “equilibration” has

occurred.

54

Table 3.1. Test for stability of measured ratios with varying sample size.
Variations were simulated by adjusting the strength of the major beam for
analysis of N20 laboratory standard gas. 211 and 3/1 refer to mlz ratios 45l44

and 46/44, respectively. SE is standard error, n is number of samples.

 

major beam (Amps) 2/1 mean SE 311 mean SE n
2.60 . 10'” 7.858 - 10“ 3.2 - 10'5 2.109 - 10“ 2.9 . 10"?—
4.43 - 10'9 7.853 . 10'3 2.0. 10'6 2.102 . 108 2.0- 10‘ 4
1.3010:8 7857-103 3010'7 2106-103 8.5- 10'7 3
1.79 - 10“ 7.858 . 10*3 4010'7 2.106 - 10*3 4.2 - 10'7 3

55

Table 3.2. Test for a memory effect during analysis. Highly enriched N20 was
prepared by mixing various quantities of labeled and unlabeled N20. Analysis
of laboratory standard N20 immediately followed each analysis of enriched
gas. Since the laboratory standard is the same as the reference gas, a value of

0 %o is expected if there is no memory effect.

date enriched sample (960) standard sample (%o)

 

1 223-103 18
1 532- 103 81
1 577 - 1o3 59
2 540.103 78
2 517 . 103 43
3 414- 103 17
3 279.103 37

56

Table 3.3. Summary of analyses of high-enrichment laboratory mixtures of
N20. 44:45:46 represents the mixing ratio of natural abundance N20, 45N20,
and “N20. Predictions '58 and 45N20 adjust for gas purity. 15a is the atom
fraction of 15N in the sample, consistent with the notation of Arah et al. (1992).
45N20 is the molecular fraction of mass 45 in the sample. Measured 15a and

“N20 are means for all samples where number of samples n is more than 1.

44:45:46 predicted 15a measured 15a predicted 45N20 measured “N20

 

1 :0: 1 0.5022 0.5057 0.0040 0.0045
1 :1 :0 0.2499 0.2658 0.4937 0.5085
1:1 :1 0.5013 0.5029 0.3283 0.3278
1 :1 :2 0.6265 0.6267 0.2460 0.2532

57

Table 3.4. Test for effects of the LiOH water trap, water in LiOH, and time on
measured enrichment of a 1:1 :2 mixture of natural abundance N20, 45N20, and
46N20. 'Hydrous' refers to hydrous LiOH used as a chemical trap, 'dried' refers
to hydrous LiOH dried on a vacuum line overnight. 45N20 and “N20 represent
the mean calculated molecular fraction for N20 of masses 45 and 46. SE is

standard error.

 

Date LiOH 45N20 SE 46N20 SE

1 dried 0.25318 0.0001 1 0.5001 1 0.00009
2 dried 0.25326 0.00026 0.49961 0.00048
2 hydrous 0.25320 0.00018 0.49980 0.00038
2 none 0.25310 0.0001 1 0.49998 0.00028

58

Table 3.5. Test for an effect of collecting sample gas quickly using pre-
evacuated vessels. Air samples were collected by diffusion into open vessels
or by rapid ﬁlling of pre-evacuated vessels when stopcocks were opened.
Samples were analyzed for N20 using the unmodiﬁed head ampliﬁer. 2/1 and
311 are mlz ratios of 45l44 and 46/44, respectively. Means 1: standard errors
are reported; n is 3. P > N is the signiﬁcance level for Student's ttest of
differences in means. Least signiﬁcant number (LSN) is the smallest number
of samples needed to demonstrate significant differences in means at a

conﬁdence level of or = 0.05.

Ratio Diffusion Pre—evac P >|t| LSN

 

2/1 7.835 - 10‘3 :l: 5.5- 10‘6 7.830 - 10‘3 i 1.4- 165 0.3965 28

3/1 2133108124- 10‘ 2129- 10311910“5 0.2602 16

59

Table 3.6. 15N enrichment of soil NOg‘ pools for laboratory denitriﬁcation
experiments, calculated and measured . Calculations are based on the
enrichments and mixing ratios of natural abundance KNOa' and highly enriched
KNOa' (99.93 atom %). Measurements use the equations of Bergsma et al.
(1999). 'n' is number of samples. Means :l: standard errors are reported for
separate experiments measuring N2 and N20. P>|t| is the signiﬁcance level for

Student's ttest.

calculated N2: measured P>|t| N20: measured P>|t| n

 

0.1032 0.1072 :1: 0.0053 0.531 0.1026 :l: 0.0003 0.210 3
0.2028 0.1969 :l: 0.0031 0.199 0.2007 :l: 0.0013 0.0336 3
0.4019 0.3827 :l: 0.0007 0.001 0.3976 i 0.0012 0.068 3

0.9993 0.9692 :l: 0.0020 0.042 (ratio out of range) 2

60

Table 3.7. Comparison of spectrometer precision by study. Precision is

reported for analysis N2 or N20. 2/1 refers to mlz 29l28 or 45l44. 3/1 refers to

30/28 or 46/44, except as noted. STD is standard deviation and n is number of

 

samples.

Citation gas ratio mean STD CV n
Siegel et al. 1982 N2 2/1 7.35 - 10" 9.3 - 10" 1.2- 10“ 15
Siegel et al. 1982 N2 3/1 *1.34- 10'5 2 7 - 10‘7 2.0- 10‘2 15
Stevens of al. 1993 N2 211 3.50 - 10'3 5.3- 10" 1.5- 10*3 7
Stevens et al. 1993 N2 3/1 1.01 - 10'5 5.3 - 10” 5.3 - 10“ 7
this study N2 2/1 7.22 - 10'3 4.8- 10'7 8.7- 10“ 12
this study N2 3/1 1.55 - 10'4 1.2- 106 7.7- 10*3 12
this study N20 211 7.83 - 10'3 2.4 - 1043 3.1 - 10“ 3
this study N20 311 2.13 - 10'3 3.3- 10‘6 1.6- 10" 3

*Siegel et al. measured 30/[29 + 28].

61

Figure 3.1. Gas puriﬁcation system connected to mass spectrometer. Valves
are conﬁgured such that during puriﬁcation, He bypasses the cold trap and
travels onto the 60 column, while sample passes into the cold trap (N2) or
through the cold trap and then to vent via the mass ﬂow controller (N20). During
analysis, helium passes through the cold trap and onto the GC column, while
the sample vessel remains open to vacuum via the mass ﬂow controller.
During N2 analysis the sample trap is not chilled. Different columns for N2 and
N20 are used. Interface helium is used only for N2 analysis. The Penning valve

passes about 10% of the gas stream to the spectrometer.

62

:58? 0%

52.50
30.38.:

 

 

MERE: ostousw

 

     

 

75:98on
$9: . oz? 559...

5.8
96.» .35 5:28
838.9395

 

.3 85?.

3.0.652
:o 9:393
a...“ lots;

) a

 

_cmm¢>
.l: :o: £88 288

P 05>Mw 0

‘l
K

 

 

 

 

 

63

Figure 3.2. Measured vs. calculated molecular fraction of 45N20 for moderate
enrichments of N20, prepared volumetrically in the laboratory from enriched

and unenriched standards; analyzed with the unmodified head ampliﬁer.

64

an ego:
c2895. 5300.9: ooze. 62930.8

mod mod cod mod No.0 5.0 oo.o
* u 4_ + 4. cod

 

i Pod

i No.0

    

w
6
B
S
n
m
D.
. 9
seamen“: zN
. - . .. O
880 x887) reed w
a.
O
m
B
I 3.5 u.
m
m.
m
.- 8.5

 

 

 

I mod

65

Figure 3.3. Isotopic character of nine N20 samples from laboratory
denitrification of prepared ‘5N03' label, plotted against the equilibrium curve. As
expected, these samples are in equilibrium. Vertical bars intersect the curve at

predicted enrichments, open circles represent measured values.

66

omd

ovd

cozom: Eoum me.

and

0N6

L

o_,.o

.1-

3 can:

cod
cod

 

 

qt—

—

A

i 3.6

r omd

J

i 00.0

1

r ovd

uonoerj JBII’IOOIOLU Xe:

l omd

 

 

cod

67

Figure 3.4. N2 and N20 ﬂuxes with cumulative precipitation for nine ﬁeld

incubations over the same highly labeled plot.

68

(uro) uouettdgoeld eAltejnwno

so 8:9“.

98
6.2.4 as} 855 of 938
So _ _ _ 58.0

ms. so oaz

 

  
    

 

5295605

8N

- Sod m
8.... u
m
m
w.
8.0 - .z
- 5.0 Mn

N

Se - ms. 3 z

 

 

 

00.0? rd

69

Figure 3. 5. Estimated soil enrichment, comparison of MS and GC, and N20

mole fraction for nine incubations represented in Figure 3.4.

70

 

 

 

   
 

no 26:.
28
85:. 829 >23 >22; 518
w 4 n \9 -914 o
0. ............................... o. .\0....¢.i°
22 + osz m oaz -r No
B
m
I to m.
a
m...
0
-- 90 w
oaz .3 2s =8 m
i w
. 9
11 m 0 W1
N2 E 29 =8 o
-- a

 

 

 

71

Chapter 4

NITROUS OXIDE MOLE FRACTION DURING DENITRIFICATION IN SOIL:
RESPONSE TO RECENT MOISTURE HISTORY VARIES AMONG
ECOSYSTEMS

Summary

Very little is known concerning the effects of recent moisture history
(antecedent moisture regime) and of ecosystem differences on the relative
proportion of N20 and N2 produced during denitriﬁcation (N20 mole fraction).
We conducted laboratory incubations of sieved soil from cropped and
successional ecosystems under two moisture histories. The soils were
pedogenically identical but had been managed differently for the past decade.
Fresh soils were air-dried, re-packed, and amended with nitrate, glucose, and
sufﬁcient water (about 85% water-ﬁlled pore space) to stimulate denitriﬁcation.
One set of incubations received 80% of prescribed water 2 d before incubation
and the remaining water at the start of the incubation; the other set of
incubations received all water at the start of the incubation. Production of
nitrous oxide and dinitrogen was estimated using acetylene inhibition
(measuring resultant N20 by gas chromatography) and also by 15N isotope
dilution (characterizing headspace samples by isotope ratio mass
spectrometry). The response of N20 mole fraction to recent moisture history
varied by ecosystem. Mean N20 mole fractions (N20 l[ N20 + N2] ) measured
using acetylene inhibition were 0.36 and 0.90 for cropped pre-wet and control

soils, respectively, and were 0.34 and 0.33 for successional pre-wet and

72

control soils. Isotope data for N20 showed that in most cases, the soil N03'
pool undergoing denitriﬁcation was nearly uniform in its N isotopic
composition. 15N isotope dilution consistently gave estimates of N2 production
that were about one-third of estimates by acetylene inhibition, suggesting that
most of the N2 came from an unlabeled source. Explicit recognition of
ecosystem differences in response of N20 mole fraction to recent moisture
history may improve modeled estimates of global N20 ﬂux.
Introduction

The proportion of denitriﬁcation end product that is nitrous oxide (N20
mole fraction) is an important aspect of the global budget of N20, a significant
greenhouse gas (IPCC 1996) and regulator of stratospheric ozone (Hahn and
Crutzen 1982). A major source of N20 is microbial denitriﬁcation in soil, which
produces dinitrogen and nitrous oxide in proportions that vary widely (Tiedje
1988, Robertson 1999). Many factors are recognized as inﬂuencing the N20
mole fraction, including soil moisture, nitrate or nitrite concentration, pH,
aeration, temperature, carbon availability, enzyme status, and moisture history
(Colbourn and Dowdell 1984, Sahrawat and Keeney 1986, Firestone and
Davidson 1989, Arah and Smith 1990, Bouwman 1990, Aulakh et al. 1992,
Hutchinson and Davidson 1993). However, few experimental studies have
considered the inﬂuence of moisture history on the nitrous oxide mole fraction
(e.g. Dendooven and Anderson 1995, Dendooven et al. 1996) and we know of
none that has looked for an interaction of moisture history and ecosystem

management history. Since most N is probably lost from soils during brief

73

periods following irrigation or rainfall (Smith and Tiedje 1979, Sextone et al.
1985, Rolston et al. 1982, Mummey et al. 1994, see also Davidson 1991 and
references therein) dependency of the N20 mole fraction on short-term soil
moisture history could have large consequences for the relationship between
nitrous oxide production and total denitriﬁcation. The study of N20 mole fraction
is hampered primarily by the difﬁculty of analyzing N2 ﬂux from soil.

In this study, we estimated nitrous oxide mole fraction for incubations of
soil from two ecosystems (row crop agriculture and early native succession
ﬁeld) and for two recent moisture histories in a factorial design. Our primary
objectives were to determine the effect of recent moisture history on N20 mole
fraction and to determine whether the effect can vary among ecosystems. A
secondary objective was to compare the use of the 1"SN isotope dilution method
with the acetylene inhibition method for estimating N20 mole fraction.

Materials and Methods
Soil collection and processing

Soils (Kalamazoo/Oshtemo soil series; Austin 1979) were collected
from the Long-term Ecological Research site at the W. K. Kellogg Biological
Station, Hickory Corners, Michigan, 42° 24' N, 85° 24' W. The soils at this site
are Typic Hapludalfs (ﬁne-loamy, mixed, mesic) derived from glacial till that
was deposited about 10,000 years ago. The ongoing LTER experiment at KBS
is a randomized complete block design with 6 replicate blocks and 7
treatments on the main site, for a total of 42 1-ha plots. We sampled three

replicates of two treatments: a high-input corn-wheat-soybean rotation and a

74

native succession treatment last plowed in 1988. The annual cropping system
is tilled and receives conventional applications of fertilizer and pesticides. The
successional treatment is managed only by occasional burning and/or removal
of woody biomass.

Soil was collected in December 1999 from blocks 4-6 on the LTER site.
For each of 6 plots (two treatments x 3 replicate blocks), 4 soil cores (2 cm
diameter by 16 cm depth) were collected at each of 5 semi-permanent
sampling stations. Soil was bulked by plot, sieved (4 mm mesh), air-dried for
several weeks (to about 1% gravimetric moisture), and stored in bags at room
temperature (“stock soil”) until the start of the experiment. Due to analytical
limitations, soil from the two ecosystems was tested on separate dates, four
weeks apart. Stock soil was tested on both dates for nitrate and ammonium
availability by KCI extraction (1M) followed by analysis using an Alpkem auto-
analyzer.

Experiment and treatments

We incubated soil from each ecosystem for 24 hours in 1 L glass
mason jars. Each jar received 150 g dry soil from one of three ﬁeld-level
replicates, packed to a volume of 125 mL (:l: ~5%) for a target bulk density of 1.2
g dry soil - cm‘a. Each jar within a replicate set was assigned to one of two
moisture histories (pre-wet or control) and one of 4 sampling strategies ('5N-
labeled soil, unlabeled soil, acetylene-amended soil, or soil for mineral N
analysis). Two additional jars were established without soil to serve as blanks

for gas analysis, for a total of 26 jars per ecosystem.

75

All soils received 9.75 mg of KN03 (about 9 pg N03'-N - g dry soil“), 20
mg glucose (about 53 pg C - g dry soil ‘1), and 56.6 mL deionized water (for a
target water-ﬁlled pore space of ~85%). “Pre-wet” soils received 80% of their
prescribed water 48 hours before the start of the incubation, with remaining
water reserved as a vector for nitrate and glucose. “Control” soils received all
of their water and nutrients as a single solution at the start of the incubation.
Blank jars received only 56.6 mL water (no soil). All soils received nitrate and
glucose immediately prior to the start of the incubation. Solutions were
delivered as a slow trickle down the edge of a tipped jar to minimize soil
disturbance and air entrapment. The delivery method produced a wetting front
that moved laterally across the soil within about 15 minutes.

The labeled soils received 9.84 mg K15N03, the molar equivalent of the
9.75 mg K“N03 received under the other three strategies. The acetylene jars
received 80 mL C2H2 at the start of the incubation for a 10% headspace
concentration, known to inhibit nitrous oxide reductase in these soils
(Robertson and Tiedje 1987). All jars were ﬁtted with air-tight lids; rubber septa
and CAJON UltraTorr unions (custom o-ring seal) were added as necessary for
syringe sampling and sampling to Pyrex vessels (0.5 L, pre-evacuated, with
stopcocks) for 15N analysis.

Sampling and analysis

The mineral-N soils were sampled destructively for analysis of nitrate

and ammonium concentrations about 2 hours after the start of the incubation

(10 g soil, dry weight equivalent, extracted in 100 mL 1M KCI). N20

76

concentrations in other jars were measured by gas chromatography at 0, 6, 12,
and 24 hours after the start of the incubation.

At the close of the incubation (24 h) gas samples were collected for
analysis by isotope ratio mass spectrometry (”N-labeled and unlabeled
treatments). The vessel stopcocks were opened for about 10 s and then
sealed. Analysis was performed within two weeks, using methods described
elsewhere (Bergsma et al. submitted). For N20, mlz ratios 46I44 and 45l44
were measured. For N2, ratios 30/28 and 29l28 were measured. Equations for
estimating the 15N enrichment of the soil mineral N pool and the fraction of
headspace gas derived from the soil mineral pool (d) require initial and ﬁnal
measurements of isotopic character (Arah 1992, Bergsma et al. 1999).
Constraints of the experiment allowed only a ﬁnal sampling. Therefore, each
labeled sample was paired with its corresponding unlabeled sample to
represent ﬁnal and initial conditions, respectively. An advantage of this pairing
is that it controls for (slight) biological and mechanical artifacts that could
inﬂuence isotopic character under the experimental conditions described
above. To guard against bias due to a label effect, N2 ﬂux was calculated as
[NZOmtyhm - Nzohbejed] for comparison to isotope data, but as [Ngomjybm -
NgOummw] for all tests of treatment effects. Differences among treatment
means were tested for statistical signiﬁcance by ANOVA, using JMPIN software

version 3.1.5 (Sall and Lehman 1996).

77

Results
Experimental design

We tested whether mineral N availability in stock successional soil
changed during the interval between the date on which cropped soil was
incubated and the date on which successional soil was incubated. Mean
extractable nitrate in stock soil from the successional ecosystem dropped
slightly between the two experimental dates, from 2.2 1 0.4 pg N - g dry soil '1 to
1.7 1 0.4 pg N - g dry soil " (statistically signiﬁcant at the 0.05 level). Mean
extractable ammonium was unchanged (15.2 1 3.2 pg N - g dry soil '1 and 15.2
1 2.0 pg N - g dry soil ").

We tested whether pre-wet and control soils differed in available mineral
N two hours after the start of incubation. Mean extractable nitrate in mineral-N
jars was 2.3 pg N - g dry soil '1 lower for pre-wet cropped soils than for controls
(19.5 1 2.5 and 21.8 1 2.2 pg N - g dry soil ’1 respectively). For successional
soils, pre-wet soils were only 0.8 pg N - g dry soil '1 lower (9.0 1 3.3 vs. 9.8 1
0.1). For both ecosystems, mean extractable ammonium was sharply higher in
pre-wet soils relative to controls: 3.1 pg N - g dry soil '1 higher for cropped soils
(4.8 1 0.4 vs. 1.7 1 0.3, respectively) and 12.7 pg N - g dry soil '1 higher for
successional soils (21.4 1 2.3 and 8.7 1 0.7).

We tested whether N20 production in labeled soils differed signiﬁcantly
from N20 production in unlabeled soils. Final N20 concentration for each
labeled jar was divided by ﬁnal N20 concentration in the corresponding

unlabeled jar. On average, the labeledzunlabeled ratio was 0.98. However, the

78

average differs by ecosystem: 1.12 for cropped soil and 0.85 for successional
soil. For this reason, production of N; by acetylene inhibition was calculated
from concentration data for unlabeled jars for tests of treatment effects, but
from labeled jars for comparison of acetylene inhibition with 15N isotope
dilution.
N20 mole fraction

The time-course of change in concentration of N20 in all jars (except
blanks, where change was negligible) is summarized in Figure 4.1. N20
accumulated more rapidly in the acetylene-amended jars than in the others.
Since acetylene inhibits the reduction of N20 to N2, production of N2 by the un-
amended soils may be inferred by difference (Yoshinari and Knowles, 1976).
Initially, total denitriﬁcation (represented by N20 produced in acetylene jars)
was less for control soil than for pre-wet soil, but was indistinguishable by the
end of the incubation. For both ecosystems, pre-wet soil produced N20
immediately and steadily throughout the incubation. Successional control soils
did not respond differently than successional pre-wet soils. However, the
response of the cropped control soils was almost identical to the response of
the cropped acetylene-amended soils, diverging only slightly by the end of the
incubation. The near-identity suggests that cropped control soils produced
signiﬁcantly less N2 than soils for the other three combinations of ecosystem
and recent moisture history.

Final N20 concentrations were used to calculate total denitriﬁcation, N20

production, and N2 production (by difference) as pg N - g dry soil’1 (Figure 4.2).

79

Results were analyzed by ANOVA. When the block effect is included in the
model, successional soils (grouped) had significantly higher total
denitriﬁcation than cropped soils, at the 0.05 conﬁdence level. Cropped
controls had signiﬁcantly higher N20 production and lower N2 production (p =
0.02 and 0.05, respectively). Although the difference in total N2 production for
successional pre-wet and successional control soils was not signiﬁcantly
different, review of time-series data for the entire incubation period shows that
pre-wet soils consistently led their corresponding controls in N2 production by a
small margin (data not shown).

Nitrous oxide mole fraction was calculated as N20 I [ N20 + N2 ] for each
combination of ecosystem, recent moisture history, and block. Analysis of the
results (ANOVA) is summarized in Table 4.1. Mean N20 mole fractions were
0.36 and 0.90 for cropped pre-wet and control soils, respectively, and were 0.34
and 0.33 for successional pre-wet and control soils. In the analysis of
variance, cropped soils showed a strong effect of recent moisture history, while
successional soils did not. The difference accounts for the highly signiﬁcant
interaction term (ecosystem x history: p = 0.012) that results when the entire
model is considered.

IsotOpic data

For N; or N20, isotopic character has been deﬁned as the relative
proportions of the three (primary) molecular fractions (2°N2, 29N2, 30N2 or 2a(N2)O,
”(N2)O, 3‘°(N2)O; see Bergsma et al., 1999). Isotopic character is conveniently

represented by a plot of 29x vs. 15a: that is, the 29(N2) molecular fraction vs. the

80

composite 15N atomic fraction. Figure 4.3 gives the isotopic character of N; at
the end of the 24 hour incubation for unlabeled (lower left) and labeled (all
other) jars. Each unlabeled/labeled pair (one line segment) represents one
incubation unit. There is strong tendency toward colinearity for pairs of line
segments representing paired pre-wet and control incubations, indicating
similar soil nitrate enrichment (consistency within replicate ﬁeld plots).
Lengths of segments representing cropped controls are generally shorter than
those for cropped pre-wet treatments, indicating smaller ﬂux. Successional
control and pre-wet segment lengths (and therefore ﬂuxes) are similar for
replicates 4 and 5, but not 6.

Figure 4.4 shows the ﬁnal isotopic character of N20 for unlabeled and
labeled jars, paired to represent incubation units. Most of the labeled
incubations resulted in equilibrium mixtures of N20 masses. Differences
between lengths of pre-wet and control segments were small.

If the soil mineral pool undergoing denitriﬁcation is not uniformly labeled,
the enrichment of the soil pool is an overestimate (Boast et al. 1988, Arah
1992, Bergsma et al. 1999). Since uniformity was not assumed in this study,
we adopt the convention of referring to the apparent enrichment of the soil
mineral pool. Apparent enrichment for each incubation is shown in Table 4.2.
For both cropped and successional soils, apparent enrichment calculated from
N2 isotope data agrees strongly with that calculated from N20 isotope data.

Enrichment of the soil mineral pool was predicted from the amount of label

81

added and the amount of nitrate initially present (Table 4.2). Predicted
enrichments agree well with the calculated enrichments.

Production of N2 estimated by isotope methods shows the same
patterns as production of N2 estimated by acetylene inhibition (Figure 4.5).
However, the isotope method gives values consistently lower than the
acetylene method. For each combination of ecosystem, moisture history, and
replicate, the ratio MSIGC was calculated, where MS refers to N2 flux by mass
spectrometry (‘SN isotope dilution method) and GC refers to N2 ﬂux by gas
chromatography using acetylene inhibition. Analysis of variance showed no
effect of block, ecosystem, or moisture history, and no interaction of ecosystem
and moisture history (P > 0.4 for all effects). WIth one outlier removed (from 12
total values), mean and standard error for MSIGC is 0.34 1 0.04.

Discussion
Experimental Design

Our experiment was designed to test for an effect of recent moisture
history (antecedent moisture regime) and for an effect of ecosystem differences
on N20 mole fraction during denitriﬁcation in soil. Validation of the design
requires that (1) tests performed on different soils were equivalent in other
respects, (2) the moisture-history treatments were equivalent in other respects,
and (3) the N20 observed was the product of denitriﬁcation.

Successional soils were tested four weeks after the cropped soils. The
very small (less than 0.5 pg N - g dry soil ’1) drop in extractable NO; -N in stock

soil during that interval, and the absence of change in NHf-N, suggest that no

82

1'32“?-

strong artifacts were introduced by the delay. The delay itself is small
compared to the initial air-dry storage time for both soils (> 8 weeks).

To test whether the two patterns of moisture addition (pre-wet vs. control)
had similar consequences for soil N status, we sampled destructively for
extractable NO; and NH4’ in the mineral-N jars as soon as possible (~2 h)
after the start of the incubation. Mean NOa'-N concentrations were 1-2 pg N - g
dry soil '1 lower for the pre-wet soils, 8 small difference relative to the amount of
N added (9 pg N - g dry soil ") and the amount of N03‘-N extracted (9-21 pg N -
g dry soil"). Extractable ammonium was sharply higher in pre-wet soils than
in control soils. However, the difference is due more to a reduction of control
[NHX], relative to background, than to the enhancement of pre-wet [NHX] (data
not shown). Apparently signiﬁcant NH; was lost (possibly through volatilization
or assimilation) when water first contacted the dry soil, but was replaced
thereafter via mineralization. However, NH: is not the substrate for
denitriﬁcation. Furthermore, since the pattern was similar across ecosystems,
this artifact does not explain the interaction of ecosystem and moisture history
in our results.

Was the observed N20 a product of denitriﬁcation? Nitrous oxide can be
produced by nitriﬁcation and dissimilatory nitrate reduction to ammonium as
well as by denitriﬁcation (T iedje 1988). However, nitriﬁcation is an aerobic
process, for which optimum moisture ranges between 30% and 70% water-
ﬁlled pore space (Davidson 1991). Many studies address the partitioning of

N20 production by source, including nitriﬁcation and denitriﬁcation (e.g.

83

Mulvaney and Kurtz 1984, Robertson and Tiedje 1987, Klemedtsson et al.
1988, Skiba et al. 1993, Mummey et al. 1994, Stevens et al. 1997, Stevens of al.
1998b, Hutsch et al. 1999, Panek et al. 2000, Wolf and Russow 2000) and also
dissimilatory nitrate reduction to ammonium (e.g. Stevens et al. 1998a). VVIth
notable exceptions (e.g. Hutchinson et al. 1993) denitriﬁcation is usually the
major source of nitrous oxide in saturated and nearly-saturated soils. Because
WFPS was about 85% for this study, we assume that N20 was a product of
denitriﬁcation. Furthermore, for a majority of the incubations, N20 derived from
soil was in isotopic equilibrium, which implicates a single, uniformly labeled
soil mineral pool (Figure 4.4; see Bergsma et al. 1999) and strongly supports
denitriﬁcation as the only important source of N20. Also, there was good
agreement between calculated and estimated enrichments for the soil NO;
pool contributing to ﬂux (Table 4.2) suggesting that N20 derived predominantly
from N032
N20 mole fraction

For the successional soil, the N20 mole fraction was about one-third.
The pre-wet treatment brought mean soil moisture to approximately 68%
WFPS, but apparently did not greatly enhance denitriﬁcation enzyme status
relative to the controls. We conclude that denitrifying enzymes, especially
nitrous oxide reductase, persisted well in the successional soil during several
months of air-dry conditions (< 1% gravimetric moisture). Enzyme persistence

in dry soils has been observed by others (e.g. Smith and Parsons, 1985).

84

For the cropped soils, however, the pre-wet treatment apparently
enhanced the activity of nitrous oxide reductase relative to the controls. N20
mole fraction was also about one-third for the pre-wet soils, but about 0.9 for
the (previously dry) control soils. Although total denitrification was Similar for
the two moisture histories, a much greater fraction of N20 was further reduced
to N2 in the pre-wet soils. We conclude that nitrous oxide reductase did not
persist well in the cropped soil when air-dry, but that its activity was signiﬁcantly
enhanced by 48 hours of high soil moisture, achieving levels similar to those
for successional soils. Since total denitriﬁcation was only slightly less for
cropped soils than for successional soils, precursor enzymes (such as nitrite
reductase and nitrate reductase) may have been less affected by drying than
was nitrous oxide reductase. Our observation that - under at least some
conditions - N20 mole fraction may be higher for the cropped soils than for the
successional soils may help explain ﬁeld data showing threefold greater
annual ﬂux of N20 from the cropped system (3.5 1 0.21 g NZO-N - ha'1 - d") than
from the successional system (1.1 1 0.05 g N20-N - ha'1 - d", Robertson et al.
2000)

To the best of our knowledge, no other published study of N20 mole
fraction has tested for a potential interaction between ecosystem effects and
moisture history effects. However, there are some reports of effects of either
ecosystem or moisture history on relative proportions of N2 and N20. Merrill
and Zak ( 1992) reported an N20 mole fraction of 0.7 to 0.9 for well-drained

sugar maple forests in northern lower Michigan; in contrast, the N20 mole

85

fraction in a silver maple - red maple swamp was 0.25. Dendooven et al.
(1996) found an effect of moisture history on relative production of nitrous oxide
and dinitrogen (N202N2) for pasture soil, but the difference was small: 0.54 for
soil cores previously submerged for 96 hours, and 0.4 for cores submerged for
6 hours. Conversion to nitrous oxide mole fraction yields values of 0.35 and
0.29: similar to the values presented here for successional soils and pre-wet
cropped soil. Mulvaney and Kurtz (1984) studied N20 and N2 flux for three 15N-
amended soils subjected to wetting and drying cycles. We calculate from their
Table 4.1 an average and standard error of 0.33 1 0.02 (n = 12), similar to the
result for our successional soils: 0.33 1 0.04 (n = 6). In a study of three N-
amended soils, Jacinthe et al. (2000) found that N20 mole fraction was initially
0.68, increased to 0.95 with imposition of a water table at a depth of 10 cm, and
decreased to 0.35 within one week.

Our results show that the dependency of nitrous oxide mole fraction on
recent moisture history can vary among ecosystems, even when the
ecosystems are pedogenically identical. The large difference in response
between the successional soils and the cropped soils may be related to
differences in soil physical properties, soil carbon patterns, and microbial
community characteristics resulting from 10 years of contrasting soil
management regimes. First, the cropped soil is plowed regularly and has
poorer aggregation than the successional soil. Because the soil used here
was sieved (4 mm), effect of aggregate structure would have been restricted to

smaller size classes of aggregates. Physical differences may have inﬂuenced

86

N20 mole fraction by altering the distribution of water and anaerobic microsites,
where most denitriﬁcation may occur (McConnaughey and Bouldin 1985).

Second, the response difference between the two soil types may be
related to differing soil carbon patterns. Although net primary productivity in the
cropped ecosystem is about double that of the successional ecosystem (mean
1 standard error = 9.24 1 1.41 vs. 4.24 :l: 0.37 MT - ha'1 - y") the successional
system is accumulating soil organic carbon while the cropped system is not
(Robertson et al. 2000). After ten years, soil organic carbon content (0 to 7.5
cm depth) was unchanged for the cropped system (1.00 1 0.05 %) but had
risen signiﬁcantly for the successional system (1.63 1 0.06 %). In addition to
greater absolute carbon content, the successional soil may have a greater
variety of substrates for microbial heterotrophs because of greater plant
species diversity. Differences in soil carbon can inﬂuence factors controlling
N20 mole fraction. Menyailo and Huwe (1999) found that 26 years of soil
development under six species of trees caused changes not only in soil
chemistry, but also in the persistence and dynamics of denitrifying enzymes.
Carbon quality (C:N ratio) was the most important soil chemical factor in
explaining differences in N20 emission among soil types.

Finally, differences in the microbial communities between the cropped
and successional ecosystems may account for the different responses to soil
moisture history. It is possible, perhaps even likely, that differing soil
properties caused functionally signiﬁcant divergence in microbial community

composition. For instance, nitrate availability is typically much lower for the

87

 

successional soil (0.63 1 0.04 pg N03'—N - g'1) than for the cropped soil (6.54 1
0.53 pg NOa'-N - g"; Robertson et al. 2000). In the successional soil, a
hypothetical sub-group of denitrifiers with the ability to maintain enzyme status
(especially NOS) during dry periods could have a competitive advantage in
exploiting the ﬂush of carbon that occurs on soil wet-up (e.g. Groffman and
Tiedje 1988), since they could use N20 as well as N03' as a terminal electron
acceptor if oxygen were limiting. ln cropped soils, the incentive for N08
maintenance would be less, because of the abundance of the more
energetically-favorable electron acceptor N031 Thus, if variation exists among
denitriﬁer taxa in their ability to maintain NOS status during soil drying, then a
putative mechanism of natural selection exists that could explain our results, in
terms of differences in microbial community composition.

Variation may indeed exist among denitriﬁers in their ability to maintain
NOS status during soil drying. Cavigelli and Robertson (2000b) isolated 31
denitriﬁer taxa from two ecosystems: the cropped ecosystem studied here and
a nearby never-tilled successional ﬁeld. They showed that considerable
variability exists among taxa for sensitivity of the NOS enzyme to varying levels
of oxygen, a parameter clearly related to soil drying. Furthermore, Cavigelli and
Robertson (2000a) found differences in denitrifying ability for whole soil
microbial communities (slurry assay) for the cropped ecosystem and the never-
tilled successional field. Denitrifying enzymes were more sensitive to oxygen
levels in the agricultural soil, and nitrous oxide reductase was more active in

the successional soil. Their results are consistent with our suggestion that the

88

microbial community in the successional soils may have experienced selection
for denitriﬁers with the ability to maintain the status of denitriﬁcation enzymes,
especially NOS.

The story of microbial community for the KBS LTER treatments is,
however, complicated. Cavigelli and Robertson (20008, 2000b) compared the
conventionally-tilled agricultural treatment (our “cropped” system) and a never-
tilled successional treatment. Our study compared the conventionally-tilled
system to a historically-tilled successional treatment. Buckley and Schmidt
(2000) used biochemical techniques to characterize relative abundance of
seven broad taxonomic groups in the microbial communities of KBS LTER
treatments. They determined that the communities from the conventionally-
tilled system and the historically-tilled system were much more similar to each
other than to the community from the never-tilled system. Therefore,
comparisons between our results and those of Cavigelli and Robertson should
be made with appropriate reserve. Still, there is a strong possibility that at finer
taxonomic levels, functionally signiﬁcant differences exist between the
microbial communities from the conventionally-tilled cropped ecosystem and
the historically-tilled successional ecosystem (D. Buckley, personal
communication). The existence of these differences and their importance for
denitriﬁcation remain to be demonstrated.

Questions of mechanism notwithstanding, evidence of a role for
moisture history in controlling N20 mole fraction has an important place in the

biogeochemistry of nitrogen. Many studies suggest that most N is lost from

89

soils during brief periods following irrigation or rainfall (Smith and Tiedje 1979,
Sextone etal. 1985, Rolston et al. 1982, Mummey et al. 1994, Davidson, 1991).
Dependency of the N20 mole fraction on short-tem soil moisture history could
have large consequences for the relationship between nitrous oxide production
and total denitriﬁcation. Since N20 ﬂux has been modeled for intervals as short
as one day (e.g. Li et al. 1992a, 1992b), the time scale implied by “short-term”
in our study (48 hours) is relevant for efforts to constrain the global N20 budget
(e.g. Bouwman 1990, Eichner 1990).
Isotope data

In principle, isotope data for N20 allow an independent estimate of N20
production. However, the estimate depends on a two-member mixing model
(Bergsma et al. 1999) in which absolute contribution from both members is
signiﬁcant. Our concentration data (Figure 4.1) show that N20 concentration
changed by almost three orders of magnitude (0.3 pg N - g dry soil " to ~200 pg
N - g dry soil '1) even in the least productive jars. We conclude that the ﬁnal
isotopic character of N20 in the headspace of labeled jars essentially
represents the isotopic character of soil-derived N20. Thus, the 15N-NZO data in
this study are suitable for tests of equilibrium and estimates of soil enrichment,
but not for independent estimates of N20 production.

Estimates of N2 production by mass spectrometry were typically only
one-third of estimates by acetylene inhibition. One of the two methods may
have been biased, or the methods may have reﬂected qualitatively different

aspects of the experimental system. The acetylene method could have been

90

biased if denitriﬁcation was enhanced by the presence of acetylene. There is
some evidence in the literature for the stimulation of denitriﬁcation by acetylene
addition (e.g. Klemedtsson et al. 1988), but usually for longer incubations and
to a lesser extent than would be the case here. Furthermore, the soils studied
here were amended with glucose, which argues against one putative
mechanism of denitriﬁcation enhancement by acetylene: namely, release from
carbon limitation (see also Topp and Germon 1986). In oxic soils acetylene
may cause denitriﬁcation to be underestimated due to scavenging of
intermediate nitric oxide (Bollman and Conrad 1997). Our soils, however, were
largely anaerobic, and the putative error (if any) is overestimation not
underestimation.

Few studies have explicitly compared the acetylene inhibition technique
with the 15N isotope dilution technique for estimation of N2 ﬂux. In their seminal
paper reviving interest in the 15N isotope dilution method for N2 ﬂux, Siegel et al.
(1982) conﬁrm accuracy of enrichment estimates for their method, but not
necessarily of ﬂux estimates. Rolston et al. (1982), measuring only N-gas
derived from fertilizer, found reasonable agreement between the acetylene
method and a 15N-accumulation method. Mosier et al. (1986) compared total
denitrification under acetylene inhibition (N20 only) with total denitriﬁcation by
15N mass spectrometry (N20 + N2; N20 catalytically reduced to N2 prior to
analysis). Acetylene-amended plots consistently gave higher ﬂuxes (nominal
MSIGC ~0.75): this was attributable to acetylene treatment itself, rather than

analytical bias between the two methods, since no difference in methods was

91

found when both were applied to samples from the acetylene-amended plots
(see Table 1, last column, in Mosier et al. 1986). Aulakh et al. (1991) reported
very similar total denitrification for two acetylene methods and for the 15N
method of Mosier et al. (1986). Arah et al. (1993) found that acetylene inhibition
consistently gave lower N2 ﬂux estimates than 15N isotope dilution, and
concluded that for heavy textured soils the acetylene block was incomplete.
Mulvaney (1988) and Mulvaney and Vanden Heuvel (1988) compared ﬂuxes of
N20 as measured by mass spectrometry (reducing N20 catalytically to N2) and
by gas chromatography. Both studies found only modest differences, and no
strong bias towards underestimation or overestimation of N20 ﬂux.

For our study, there is no strong justiﬁcation for questioning the accuracy
the N2 measurements made by acetylene inhibition. The difference between
the two methods is perhaps explained as bias in the 15N-based
measurements — an explanation that is not entirely satisfying.

Underestimation is expected from the 15N-dilution technique when the soil
mineral N pool undergoing denitrification is not uniform (Boast et al. 1988, Arah
1992, Bergsma et al. 1999). But how much underestimation? Simulation
(Arah 1992) and direct calculation (Bergsma et al. 1999) show that if ﬂux
derives from a very large number of pools with enrichments randomly
distributed from natural abundance levels to 100%, the central tendency of the
underestimation is 0.75. Albeit theoretical, the value 0.75 serves as a
convenient null hypothesis. It does not explain our nominal MSIGC ratio of

0.34. Also, Figure 4.4 shows that for most incubations, N20 derived from soil

92

was probably in equilibrium, implying a well-mixed soil source. Since N20 is
the direct precursor of N2 (Payne 1981), one would expect N; from soil also to
be in equilibrium (Focht 1985) and therefore free of the underestimation
ascribed to non—uniform pools. Furthermore, estimates of enrichment (of the
soil mineral pool undergoing denitrification) based on N2 in our study agreed
well with estimates based on N20. Others have found similar results (Mulvaney
and Kurtz 1984, Mosier et al. 1986).

Given the above considerations, the most satisfying explanation for the
differences between the two methods of calculating N2 production is that the
methods gave qualitatively different information about the observed system:
the acetylene method reported gross N2 production while the 15N dilution
method reported only production from a highly enriched, uniformly labeled pool,
i.e. the enriched N20 or its substrate. A second, unenriched soil mineral N pool
was also a source of N2 (indeed the major source), but not a net source of N20.
Under these circumstances, N2 production would have been underestimated
without affecting the estimate of enrichment for the labeled pool (see Focht
1985) which would explain the agreement of N2 and N20 data for estimates of
pool enrichment (Table 4.2). Estimates of enrichment (of the soil mineral pool)
based on isotopic data also agreed well with estimates predicted by mass
balance (based on knowledge of extractable NO;{ levels for stock soil and
knowledge of the magnitude of the K15N03 addition; Table 4.2). The agreement
of the mass balance estimates and the 15N estimates suggests further that the

putative unlabeled source of N2 is not a static, extractable NOg’ pool. Perhaps

93

the unlabeled source is a dynamic result of mineralization and nitrification
occurring as a consequence of rapid soil wet-up, tightly-coupled to
denitriﬁcation. This scenario implies nitriﬁcation rates (a few pg N - g dry soil'1 -
d") that are about an order of magnitude greater than typical potential
nitriﬁcation rates for these soils in the ﬁeld (about 0.1 pg N - g dry soil’1 - d";
Robertson et al. 2000). More work is needed to account for the interesting
differences between the acetylene inhibition method and the 15N-dilution
method for estimating N2 ﬂux as used in this study.
Conclusions

The design of the study as implemented was suitable to test for effects
of recent moisture history (antecedent moisture regime) and ecosystem
differences on N20 mole fraction during denitriﬁcation. N20 mole fraction in
successional soils was not affected by moisture history, but in cropped soils it
was sharply lower when soil moisture had been high for 48 hours prior to
incubation. We suggest that persistence of nitrous oxide reductase in the
successional soils was less sensitive to water stress during soil drying,
perhaps because the lower level of native soil nitrate selects (in the
successional microbial community) for denitriﬁer taxa with enhanced capacity
for enzyme maintenance. Explicit recognition of ecosystem differences in
response of N20 mole fraction to recent moisture history may improve modeled
estimates of global N20 ﬂux. Furthermore, understanding the impact of soil
management regimes on mole fraction dynamics within ecosystems may lead

to strategies that minimize ﬂux of N20 to the atmosphere.

94

The isotope data show that the mineral N pool undergoing denitriﬁcation
was isotopically uniform in most cases. There is no conclusive explanation for
the strong differences between estimates of N2 production by acetylene
inhibition and by 15N isotope dilution. Perhaps, under the experimental
conditions described, there existed an alternative substrate for production of N2,
but not for N20; the 15N method may then have reported only production of
labeled N2, while acetylene inhibition would have reﬂected gross production of
N2. Factors likely contributing to the difference between methods are the
complexity of the soil environment and the dynamic nature of N transformations

during rapid re-wetting of soil.

95

Table 4.1. Nitrous oxide mole fraction ( N20 I [ N20 + N2] ) analyzed by
ecosystem and recent soil moisture history. Estimates are mean 1 standard
error, in pg N - g dry soil". Effects are P values (Prob. > F). “Ecosystem by

history” is the interaction term.

96

 

«N50 mmood 83.0 mvvod mod N mvd - - uoEnEoo
- - SE6 ommod - no.0 a and 9.0 a wad .m:o_mmoooam
- - R85 88.0 - a; a 8.0 mod n 8.0 8:88
{on to ice be foo be
Ex“: $31 $de $21 929: 929: 929:
boE: 3 Soto Soto Soto come :36 :moE
823388 82338 >632 3.83 605988 .228 103.05 Ecum>moom

._..v 2an

97

Table 4.2. Comparison, by ecosystem and replicate, of predicted soil N03'

enrichment (atom fraction 15N) with apparent enrichment of the soil pool

undergoing denitriﬁcation. “Predicted” is calculated by mass balance from the

extractable NO3' levels in stock soil and the known addition of KNOa. Apparent

enrichments are from N2 isotope data (“by N2”), from N20 isotope data (“by

N20”), or “average” of N2 and N20 estimates.

 

Ecosystem Rep. predicted by N2 by N20 average
cropped 4 0.42 0.57 0.53 0.55
cropped 5 0.35 0.31 0.38 0.34
cropped 6 0.42 0.71 0.58 0.64
successional 4 0.90 0.87 0.87 0.87
successional 5 0.78 0.83 0.83 0.83
successional 6 0.84 0.84 0.86 0.85

98

Figure 4.1. Change in headspace concentrations (volume/volume) of nitrous
oxide during incubation, with selected (for clarity) standard error bars. Dashed
lines = successional soils, solid lines = cropped soils. Open symbols =
control, ﬁlled symbols = pre-wet. Gray squares = + acetylene, black diamonds
= - acetylene. Control and pre-wet curves diverge strongly for cropped soils
(solid lines with open or ﬁlled diamonds) but not for successional soils

(dashed lines with open or filled diamonds).

99

 

.3. 2:9:

5:335 co Ego:

 

 

  

vm o
4 919°
-- Two...
.. v-moN
H. -- Eon. w
.m
-- "two.“‘ 0
. w
61:8 .8390 IT Tmom W
go‘s-ca condo-.0 IT . W
. _o._Eoo ._m:o_wmooo:m :6.-- -- Ymoo x
.. woes-ca ._m:o_mmooo:m :6.-- m
8:038"... + 63:8 .8380 limit -- Ymos ...
a . .. 05.38.... + “oz-ma condo-6 .......I.......
0:238er .9250 ._m:o_mmooo:m :-E.-.. -. Twod
.. 820581-6305 ._m:o_mmoooam ----l.
H. -- 1mg

 

 

Wmoé

100

Figure 4.2. Summary of production of N2 and N20. “Total denitriﬁcation” is N20
+ N;- by acetylene inhibition, “N20” is production of N20 in the absence of
acetylene, and “N2” is the difference. Even though total denitriﬁcation did not
differ signiﬁcantly between moisture histories for cropped soils (bars with solid
borders), the control incubations produced signiﬁcantly more N20 and
correspondingly less N2. Bars with dashed borders represent soils from

successional plots.

101

Nz ONZ cosmoEono .20...

 

  

.958 ._mco_mmo8:m F 83.08 ._m:o_mmm83m l

61:8 cacao-6 W 83-05 .8808 I

as 2:9“.

c'>.
v
nos Mp 6 - N 6n

 

 

102

Figure 4.3. Isotopic character of N2 for labeled and unlabeled jars at the end of
the incubation, paired (by line segment) to represent incubation units. Dashed
lines = successional soils, solid lines = cropped soils. Open symbols =
controls, ﬁlled symbols = pre-wet. Circles = Replicate 4, squares = Replicate
5, triangles = Replicate 6. Due to the ovenrvhelming abundance of unlabeled
N2 from the atmosphere relative to labeled soil—derived N2, displacement of
isotopic character during the incubation (i.e., length of the line segments) is
much smaller than for N20 (Figure 4.4). Only a small portion of the equilibrium
curve (downward-opening parabola in Figure 4.4) is visible at this scale. For
cropped soils, control segments (solid lines with open symbols) are much
shorter than pre-wet segments (solid lines with closed symbols) indicating
less N2 production for controls. For successional soils (dashed lines) the

opposite effect or no effect is seen.

103

:ozomt Eoﬁ 2....

m-mmm.m m-mom.m mummmd m-wow.m m-mmk

.m m-mond

.2. 28:.

Wmmod

 

 

_
H A _ d

cos-ma ”285.?- 85...
63:8 ”£88? :80
.moﬁoiom 2 8888 88% .856

)\
tine
X».
\ 0
“\“ﬂst
C 0“”9“

. Q
s‘!“‘ .0“
8 Q‘ ‘ C
S 8. ‘ .
\ ‘ ’
a. \5
- .
9‘ ‘0‘
\ \\ $9
‘ “
N O
8‘ ‘ . ‘
\
8 ‘ ‘-
9‘ \‘ ‘8‘
\
t \s .1 s
\.\ \. s Q
‘ ‘
s I. \.w1 \
I‘ .9
9 g C. o \
’
‘ - Q ‘
\ .
.. 1... l .
8 ‘
5‘ ‘ Q
‘n' 8‘ ‘
t \ \
o I Q
E
\( 5‘ \‘
3‘ ‘0 ‘
.1 9 S
\\ 5‘ 0
00‘. 1 \‘
‘ \
3 3‘ ‘
’ b.\ ‘
9. \ ‘
$3 \
\ \
OI \\
5
\
Q- s
‘ I
-- 8890 .moE. 9.8
‘ I I
Q
‘
\
‘
‘
\
\
‘
‘
Q
Q
‘
‘
‘

.m:o_mmm8:m “mos. 88%

l

l

1

l

m-momN

- mummmk

. m-mva

104

- m-moms

uonoeu Jemoejour ‘Nez

- m-mmmN

- mimovﬁ

 

 

m-mmvN

Figure 4.4. Isotopic character of N20 for labeled and unlabeled jars at the end
of the incubation, paired (by line segment) to represent incubation units.
Dashed lines = successional soils, solid lines = cropped soils. Open symbols
= controls, filled symbols = pre-wet. Circles = Replicate 4, squares = Replicate
5, triangles = Replicate 6. Due to the overwhelming abundance of N20 from
soil relative to unlabeled atmospheric N20, enriched samples represent
essentially the isotopic character of N20 derived from soil. Samples falling on
or near the equilibrium curve (downward-opening parabola) indicate a source
in isotopic equilibrium, implying a single, uniformly labeled substrate pool (e.g.

a homogenous mixture of native N03' and 15N03' label).

105

00;

 

828.30”. 2 0:888 wedge 6956

5.8:: 2:85 29

.3. 239....

00.0 00.0 0v.0 0N.0 00.0

p p —
H _ _

cos-ca 8.8:...» 8...... hams-«ea
61:8 ”2853 :80

.-

 

‘
I!
‘0
I I
“. Is.
8‘ 8 I.
9 C \
‘0 I... 1‘
‘I {.6 S
‘ ‘
0 1 1.
‘ ‘ ‘
“ .I‘ ‘n- ’
“‘ 0‘ ‘ ’ﬂ‘
‘ C \
‘ ‘ ‘ “
I 0 1 19
“‘ “ ‘8 E‘
.‘ ‘. “ “.
0 \ 6 I
‘ ‘ “
9 6 6 I
‘0 C \ C O
I. .1 1 5 O
“ 0‘ ‘n‘. 0 ‘
11 t 1 1 O
0 .1 o. 0 ‘
“ C.‘ ‘ .
1 10
9‘ .1 c. 5
I 11 1 1 O
I. a. 9 s O
| s 1. O
1 1 t 10
‘ 0‘ ‘
$ 6 .I Q
\s 1.. 1 c
\ i b. .
1 1 t c
\1 s. u. 0
‘ " ‘
11 t 1 v
1 $6 ‘9‘.
vs 1 o
1 1 c
1 1 o
.1 1 c
s O
9 C
.u U
0‘ "
‘ ‘
i .
\ .
.1 i
.1 ’
I. .V
. .
3 ‘

53.5...38 .028

   

L

-0_..0

63

-- 09.0)

g.
N
0
NW

I

I
m
N
o

1

T
o
‘8
o

-- 00.0

uoltoerj rajnoejour

l

- 0Y0

-- 0Y0

 

00.0

 

106

Figure 4.5. N: production calculated independently by the acetylene inhibition
technique (AIT) and by mass spectrometry (M8), for both cropped and
successional soils. In every case but one, MS gave a much smaller estimate

of N2 production than did AIT.

107

 

.3 2:9“.

w: ._mco_mmooo:m t< __m:o_mmmoo:m ms. 62520 ._._< .3390
it 0. Fl
v.0.
1' 0.0
l o._‘

a,
N
l_|go's Mp 6 - N 6n

 

l
C!
00

  

65:8 cm I
6235 my. I
.9550 mm l
83.95 mm E
6:50 5. a
63.05 vm n

 

 

 

 

108

OVERALL CONCLUSION

A general conclusion that emerges from my research is that progress in
understanding the relationship between denitrification and N20 flux from soil is
still methods-limited. The problem arises from the position that denitriﬁcation
occupies in the nitrogen cycle: right on the brink of a sea of molecular N that
constantly scours the beaches of inquiry. Acetylene inhibition and 15N dilution
are primarily devices to increase the sensitivity of measuring N2 production, the
missing piece of the puzzle. Unfortunately, both interact with the system they
were designed to explore, thereby compromising the interpretation. I do not
mean to suggest that accurate measurements of N2 are rare or impossible,
only that they are difﬁcult and far from routine.

The good news is that when the tide goes out, the stories we ﬁnd on the
beach tend to be consistent. Variation in N2 and N20 production seems to
abide proximally by principles of differential enzyme induction, and more distally
by consistently-recognized ecological controls. The main question for the
biogeochemistry community is, “How much needs to be known about
ecological controls -- as distributed across ecosystems - in order to
adequately predict regional and global N20 flux?” When the practical threshold
for data gathering meets that need, a major breakthrough will be within reach. I
hope that in some way my brief dashes among the waves will have made a

contribution.

109

 

 

APPENDIX

110

APPENDIX

111

Table A1. Tests of significance (of differences among means) for Figure 4.2:
probabilities of ﬁnding the observed differences in means by chance alone.
Each row represents a single model of “Response”, incorporating the effects
for which there are column entries. In most cases, models returning P > 0.05
for an effect were re-run without that effect (“-”). Dropping the effect (“-") is a
notational convenience only; statistically it is no different from grouping the

responses across levels (“grouped”). Levels of block are [4, 5, 6]. Levels of

ecosystem are [cropped, successional]. Levels of history are [pre-wet, control].

“NA” means the effect (column) is not meaningful for testing the response

(row).

112

.7...

x " 0'4. mugs}? 44.4

 

 

 

<2 whovd noaaoco $3.65 ”2983005 ouz

- m8; - 8:385 ouz

82.0 - - 8:885 owz

88.0 88d - 8:385 onz

88.0 983 88.0 8:885 ouz

- <2 83.0 .mco_mmmoo:m ”cozoaooa ouz

89.0 <2 <83 385883 “8.6385 omz

- <2 55.0 8&20 ”88:85 onz

88.0 <2 88d 8&90 ”83:85 omz

- - £86 58.? 8:852ch :29

- 59.0 88.0 686 8:852ch .92
83.0 moomd 88.0 88.0 8:85:38 .92
>552 x x003 >8sz Emumxmoom xooE omcoamom

.3. 2an

113

 

 

 

E

2

v00 F .0

3.00.0

08.5.0

803.0

03.0.0

3.00.0

nmmod

vmmod

05:30

mmood

0000.0

00.0.0

0050.0

53.20 ”20003020 Nz
$3.20 ”2283005 Nz
$223808 ”00003090 Nz
5228803 ”20003020 ANz
08090 E26390 ANz
cozozpoa Nz

:26:an ANz

2.835580 2 2%.;

114

LITERATURE CITED

115

 

Literature Cited

Arah, J. R. M. 1992. New formulae for mass spectrometric analysis of nitrous
oxide and dinitrogen emissions. Soil Science Society of America Journal
56: 795-800.

Arah, J. R. M. 1997. Apportioning nitrous oxide ﬂuxes between nitriﬁcation and
denitriﬁcation using gas-phase mass spectrometry. Soil Biology and
Biochemistry 29: 1295-1299.

Arah, J. R. M. and K. A. Smith. 1990. Factors inﬂuencing the fraction of the
gaseous products of soil denitriﬁcation evolved to the atmosphere as
nitrous oxide. Pages 475-480 in A F. Bouwman (ed.) Soils and the
Greenhouse Effect. John Wiley and Sons, Ltd.

Arah, J. R. M., I. J Crichton, and K A. Smith. 1993. Denitriﬁcation measured

 

directly using a single-inlet mass spectrometer and by acetylene
inhibition. Soil Biology and Biochemistry 25: 233-238.

Arnold, P. W. 1954. Losses of nitrous oxide from soil. Journal of Soil Science
5: 116-128.

Aulakh, M. S., J. W. Doran, and A. R. Mosier. 1992. Soil denitriﬁcation -
significance, measurement, and effects of management. Advances in
Soil Science 18: 1-57.

Aulakh, M.S., J. W. Doran, and A. R. Mosier. 1991. Field evaluation offour
methods for measuring denitriﬁcation. Soil Science Society of America

Journal 55: 1332-1 338.

116

is"!

Austin, Franklin R. 1979. Soil Survey of Kalamazoo County, Michigan. United
States Department of Agriculture, Soil Conservation Service, in
cooperation with Michigan Agricultural Experiment Station. The Soil
Conservation Service: Washington.

Bergsma, T. T., N. E. Ostrom, M. Emmons, and G. P. Robertson. (submitted).
High-sensitivity determination of nitrous oxide ﬂux from nitrogen-15-
labeled soil by isotope ratio mass spectrometry. Soil Science Society of
America Journal.

Bergsma, T. T., Q. C. Bergsma, N. E. Ostrom, and G. P. Robertson. 1999. A
heuristic model for the calculation of dinitrogen and nitrous oxide ﬂux
from nitrogen-15-Iabeled soil. Soil Science Society of America Journal
63: 1709-1716.

Betlach, M. R. and J. M. Tiedje. 1981. Kinetic explanation for accumulation of
nitrite, nitric oxide, and nitrous oxide during bacterial denitriﬁcation.
Applied and Environmental Microbiology 42: 1074-1084.

Blackmer, A. M. and J. M Bremner. 1978. Inhibitory effect of nitrate on reduction
of N20 to N2 by soil microorganisms. Soil Biology and Biochemistry 10:
187-191.

Blackmer, A. M. and J. M. Bremner. 1979. Stimulatory effect of nitrate on
reduction of N20 to N2 by soil microorganisms. Soil Biology and

Biochemistry 1 1: 313-315.

117

 

Boast, C. W., R. L. Mulvaney, and P. Baveye. 1988. Evaluation of nitrogen-15
tracer techniques for direct measurement of denitriﬁcation in soil: 1.
Theory. Soil Science Society of America Journal 52: 1317-1322.

Bollman, A. and R. Conrad. 1997. Acetylene blockage technique leads to
underestimation of denitriﬁcation rates in oxic soils due to scavenging of
intermediate nitric oxide. Soil Biology and Biochemistry 29: 1067-1077.

Bouwman, A. F. 1990. Exchange of greenhouse gases between terrestrial

_ ecosystems and the atmosphere. Pages 61-127 in A. F. Bouwman (ed.)
Soils and the greenhouse effect. John Wiley 8. Sons, New York.

Boyd, S. R., A. Rejou-Michel, and M Javoy. 1994. Non-cryogenic puriﬁcation of
nanomole quantities of nitrogen gas for isotopic analysis. Analytical
Chemistry 66: 1396-1402.

Brooks, P. 0., D. J. Herman, G. J. Atkins, S. J. Prosser, and A. Barrie. 1993.
Rapid, isotopic analysis of selected soil gases at atmospheric
concentrations. Pages 193-202 in Agricultural Ecosystem Effects on
Trace Gases and Global Climate Change. ASA Special Publication no.
55. American Society of Agronomy, Madison, WI, USA.

Buckley, 0. H. and T. M. Schmidt. 2000. The structure of microbial
communities in soil and the lasting impact of cultivation. Microbial
Ecology (in press).

Cavigelli, M. A. and G. P. Robertson. 2000a. The functional signiﬁcance of
denitriﬁer community composition in a terrestrial ecosystem. Ecology

81: 1402-1414.

118

.P

 

Cavigelli, M. A and G. P. Robertson. 2000b. The role of denitriﬁer diversity in
rates of nitrous oxide consumption in a terrestrial ecosystem. Soil
Biology and Biochemistry (in press).

Colboum P. and R. J. Dowdell. 1984. Denitriﬁcation in ﬁeld soils. Plant and
Soil 76: 213-226.

Cooper, G. S. and R. L. Smith. 1963. Sequence of products formed during
denitriﬁcation in some diverse western soils. Soil Science Society of
America Proceedings 27: 659-662.

Davidson, E. A. 1991. Fluxes of nitrous oxide and nitric oxide from terrestrial
ecosystems. Pages 219-235 in John E. Rogers, William B. Wnitman
(eds.) Microbial production and consumption of greenhouse gases:
methane, nitrous oxides, and halomethanes. American Society for
Microbiology, Washington, D. 0, USA.

Dendooven, L. and J. M. Anderson. 1994. Dynamics of reduction enzymes
involved in the denitriﬁcation process in pasture soil. Soil Biology and
Biochemistry 26: 1501 -1 506.

Dendooven, L. and J. M. Anderson. 1995. Maintenance of denitriﬁcation
potential in pasture soil following anaerobic events. Soil Biology and
Biochemistry 27: 1251 -1 260.

Dendooven, L., L. Duchateau, and J. M. Anderson. 1996b. Gaseous products
of the denitriﬁcation process as affected by the antecedent water regime

of the soil. Soil Biology and Biochemistry 28: 239-245.

119

EA? -
I

1.2 _

 

Dendooven, L., P. Splatt, and J. M. Anderson. 1996a. Denitriﬁcation in
permanent pasture soil as affected by different forms of C substrate.
Soil Biology and Biochemistry 28: 141-149.

Dendooven, L., P. Splatt, J. M. Anderson, and D. Scholeﬁeld. 1994. Kinetics of
the denitriﬁcation process in a soil under permanent pasture. Soil
Biology and Biochemistry 26: 361 -370.

Eichner, M. J. 1990. Nitrous oxide from fertilized soils: summary of available
data. Journal of Environmental Quality 19: 272-280.

Firestone, M. K. and E. A Davidson. 1989. Microbiological basis of NO and
N20 production and consumption in soil. Pages 7-21 in M. 0. Andreae
and D. S. Schimel (eds) Exchange of Trace Gases between Terrestrial
Ecosystems and the Atmosphere. John Wiley and Sons: Chichester;

New York.

 

Firestone, M. K. and J. M. Tiedje. 1979. Temporal change in nitrous oxide and
dinitrogen from denitrification following onset of anaerobiosis. Applied
and Environmental Microbiology 38: 673-679.

Firestone, M. K., M. S. Smith, R. B. Firestone, and J. M. Tiedje. 1979. The
inﬂuence of nitrate, nitrite, and oxygen on the composition of the
gaseous products of denitrification in soil. Soil Science Society of
America Journal 43: 1140-1144.

Focht, D. D. 1974. The effect of temperature, pH, and aeration on the
production of nitrous oxide and gaseous nitrogen - a zero-order model.

Soil Science 1 18: 173-179.

120

Focht, D. D. 1985. Differences in nitrogen-15 enrichments of evolved nitrous
oxide and dinitrogen and the question of a uniform nitrate-15 pool. Soil
Science Society of America Journal 49: 786-787.

Gaskell, J. F., A. M. Blackmer and J. M. Bremner. 1981. Comparison of effects
of nitrate, nitrite and nitric oxide on reduction of nitrous oxide to dinitrogen
by soil microorganisms. Soil Science Society of America Journal 45:

1 124-1 127.

Groffman, P. M. and J. M. Tiedje. 1988. Denitriﬁcation hysteresis during wetting
and drying cycles in soil. Soil Science Society of America Journal 52:
1626-1629.

Hahn, J. and P. J. Crutzen. 1982. The role of ﬁxed nitrogen in atmospheric
photochemistry. Philosophical Transactions of the Royal Society of
London 8296: 521-541.

Hallmark, S. L. and R. E. Terry. 1985. Field measurement of denitriﬁcation in
irrigated soils. Soil Science 140: 35-44.

Hart, S. C., J. M. Stark, E. A. Davidson, and M. K Firestone. 1994. Nitrogen
mineralization, immobilization, and nitriﬁcation. Pages 985-1018 in R.
W. Weaver et al. (eds) Methods of Soil Analysis, Part 2: Microbiological
and Biochemical Properties-SSSA Book Series, no. 5.

Hauck, R. D., and D. R. Bouldin. 1961. Distribution of isotopic nitrogen in

nitrogen gas during denitriﬁcation. Nature (London) 191: 871-872.

121

 

Hauck, R. D., S. W. Melsted, and P. E. Yankwich. 1958. Use of N-isotope
distribution in nitrogen gas in the study of denitriﬁcation. Soil Science
86: 287-291.

Healy, W. R., R. G. Striegl, T. F. Russell, G. L. Hutchinson, and G. P. Livingston.
1996. Numerical evaluation of static-chamber measurements of soil-
atmosphere gas exchange: identiﬁcation of physical processes. Soil
Science Society of America Journal 60: 740-747.

Howarth R. W, G. Billen, D. Swaney, A. Townsend, N. Jaworski, K. Lajtha, J. A.
Downing, R. N. Elmgren, T. Jordan, F. Berendse, J. Freney, V. Kudeyarov,
P. Murdoch, Z. L. Zhu. 1996. Regional nitrogen budgets and riverine N
and P ﬂuxes for the drainages to the North-Atlantic Ocean - natural and
human influences. Biogeochemistry 35: 75-139.

Hutchinson, G. L. and E. A. Davidson. 1993. Processes for production and
consumption of gaseous nitrous oxide in soil. Pages 79-94 in G. A.
Peterson, P. S. Baenziger (eds.-in-chief) Agricultural Ecosystem effects
on Trace Gases and Global Climate Change. ASA Special Publication
no. 55.

Hutchinson, G. L., W. D. Guenzi, and G. P. Livingston. 1993. Soil water controls
on aerobic soil emission of gaseous nitrogen oxides. Soil Biology and
Biochemistry 25: 1-9.

Hutsch, B. W., X. Z. Wang, K. Feng, F. Yan, and S. Schubert. 1999. Zeitschn'ft

fur Pflanzenernahrung und Bodenkunde 162: 607-613.

122

 

lntergovemmental Panel on Climate Change. 1996. J. T. Houghton, L. G.
Meira Filho, B. A. Callander, N. Harris, A. Kattenberg, and K. Maskell
(ed.), Climate Change 1995: The Science of Climate Change.
Cambridge University Press, Cambridge.

Jacinthe, P. A., W. A. Dick, L. C. Brown. 2000. Bioremediation of nitrate-
contaminated shallow soils and waters via water table management
techniques: evolution and release of nitrous oxide. Soil Biology and
Biochemistry 32: 371 -382.

Kaspar, H. F. and J. M. Tiedje. 1994. Anaerobic bacteria and processes.
Pages 223-243 in R. W. Weaver et al. (eds) Methods of Soil Analysis,
Part 2: Microbiological and Biochemical Properties-SSSA Book Series,
no. 5.

Klemedtsson, L., B. H Svensson, and T. Rosswall. 1988. Relationships
between soil moisture content and nitrous oxide production during
nitriﬁcation and denitriﬁcation. Biology and Fertility of Soils 6: 106-111.

Koskinen, W. C. and D. R. Keeney. 1982. Effect of pH on the rate of gaseous
production of denitriﬁcation in a silt loam soil. Soil Science Society of
America Journal 46: 1 165-1 167.

Kroeze, C., A. Mosier, and L. Bouwman. 1999. Closing the global N20 budget:
A retrospective analysis 1500-1994. Global Biogeochemical Cycles 13:
1-8.

123

 

 

Lensi, R. and A. Chalamet. 1982. Denitriﬁcation in waterlogged soils: in situ
temperature-dependent variations. Soil Biology and Biochemistry 14:

1 51 -1 55.

Letey, J., N. Valoras, A. Hadas, and D. D. Focht. 1980a. Effect of air-ﬁlled
porosity, nitrate concentration, and time on the ratio of ~20/N2 evolution
during denitriﬁcation. Journal of Environmental Quality 9: 227-231.

Letey, J., W. A. Jury, A. Hadas, and N. Valoras. 1980b. Gas diffusion as a factor
in laboratory incubation studies on denitriﬁcation. Journal of
Environmental Quality 9: 223-227.

Li, C., S. Frolking, and T. A. Frolking. 1992a. A model of nitrous oxide evolution
from soil driven by rainfall events: 1. Model structure and sensitivity.
Journal of Geophysical Research 97: 9757-9776.

Li, C., S. Frolking, and T. A. Frolking. 1992b. A model of nitrous oxide evolution
from soil driven by rainfall events: 2. Model applications. Journal of
Geophysical Research 97: 9777-9783.

Linn, D. M. and J. W. Doran. 1984. Effect of water-ﬁlled pore space on carbon
dioxide and nitrous oxide production in tilled and non-tilled soils. Soil
Science Society of America Journal 48: 1267-1272.

Matthews, E. 1994. Nitrogenous fertilizers - global distribution of consumption
and associated emissions of nitrous oxide and ammonia. Global

Biogeochemical Cycles 8: 411-439.

124

McConnaughey, P. K. and D. R. Bouldin. 1985. Transient Microsite Models of
Denitriﬁcation: I. Model Development. Soil Science Society of America
Journal 49: 886-891.

McKenney, D. J., C. F. Drury, W. I. Findlay, B. Mutus, T. McDonnell and C. Gajda.
1994. Kinetics of denitriﬁcation by Pseudomonas ﬂuorescens: oxygen
effects. Soil Biology and Biochemistry 26: 901 -908.

Menyailo, O. V. and B. Huwe. 1999. Activity of denitriﬁcation and dynamics of
N20 release in soils under six tree species and grassland in central
Siberia. Zeitschrift fur Pﬂanzenernahrung und Bodenkunde 162: 533-
538.

Merrill, A. G. and D. R. Zak. 1992. Factors controlling denitriﬁcation rates in
upland and swamp forests. Canadian Journal of Forest Research 22:

1 597-1604.

Mosier, A. R. 1989. Chamber and isotope techniques. Pages 175-187 in M. O.
Andreae and D. S. Schimel (eds) Exchange of Trace Gases between
Terrestrial Ecosystems and the Atmosphere. John Wiley and Sons:
Chichester; New York.

Mosier, A. R. and L. Klemedtsson. 1994. Measuring denitriﬁcation in the ﬁeld.
Pages 1047-1066 in R. W. Weaver et al. (eds) Methods of Soil Analysis,
Part 2: Microbiological and Biochemical Properties-SSSA Book Series,

no. 5.

125

Mosier, A. R., S. K Mohanty, A. Bhadrachalam, and S. P. Chakravorti. 1990.
Evolution of dinitrogen and nitrous oxide from the soil to the atmosphere
through rice plants. Biology and Fertility of Soils 9: 61-67.

Mosier, A. R., W. D. Guenzi, and E. E. Schweizer. 1986a. Soil losses of
dinitrogen and nitrous oxide from irrigated crops in northeastern
Colorado. Soil Science Society of America Journal 50: 344-348.

Mosier, A. R., W. D. Guenzi, and E. E. Schweizer. 1986b. Field denitriﬁcation
estimation by nitrogen-15 and acetylene inhibition techniques. Soil
Science Society of America Journal 50: 831-833.

Mosier, A., C. Kroeze, C. Nevison, O. Oenema, S. Seitzinger, 0. van Cleemput.
1998. Closing the global N20 budget: nitrous oxide emissions through
the agricultural nitrogen cycle - OECD/IPCCIIEA phase II development of
IPCC guidelines for national greenhouse gas inventory methodology.
Nutrient Cycling in Agroecosystems 52: 225-248.

Mulvaney, R. L. 1984. Determination of 1SN-Iabeled dinitrogen and nitrous
oxide with triple collector mass spectrometers. Soil Science Society of
America Journal 48: 690-692.

Mulvaney, R. L. 1988. Evaluation of nitrogen-15 tracer techniques for direct
measurement of denitriﬁcation in soil: lll. Laboratory studies. Soil
Science Society of America Journal 52: 1327-1332.

Mulvaney, R. L. and C. W. Boast. 1986. Equations for determination of
nitrogen—15 labeled dinitrogen and nitrous oxide by mass spectrometry.

Soil Science Society of America Journal 50: 360-363.

126

 

Mulvaney, R. L. and L. T. Kurtz. 1982. A new method for determination of 15N-
labeled nitrous oxide. Soil Science Society of America Journal 46: 1178-
1 184.

Mulvaney, R. L. and L. T. Kurtz. 1984. Evolution of dinitrogen and nitrous oxide
from nitrogen-15 fertilized soil cores subjected to wetting and drying
cycles. Soil Science Society of America Journal 48: 596-602.

Mulvaney, R. L. and L. T. Kurtz. 1985. Reply to “Differences in nitrogen-15
enrichments of evolved nitrous oxide and dinitrogen and the question of
a uniform nitrate-15 pool”. Soil Science Society of America Journal 49:
787.

Mulvaney, R. L. and R. M. Vanden Heuvel. 1988. Evaluation of nitrogen-15
tracer techniques for direct measurement of denitriﬁcation in soil: IV.
Field studies. Soil Science Society of America Journal 52: 1332-1337.

Mummey, D. L., J. L. Smith, and H. Bolton. 1994. Nitrous oxide ﬂux from a
shrub-steppe ecosystem: sources and regulation. Soil Biology and
Biochemistry 26: 279-286.

Nommik, H. 1956. Investigation on denitrification in soil. Acata Agriculturae
Scandinavica 6: 195-228.

Panek, J. A., P. A Matson, l. Ortiz-Monasterio, and P. Brooks. 2000.
Distinguishing nitriﬁcation and denitrification sources of N20 in a
Mexican wheat system using N-15. Ecological Applications 10: 506-514.

Paul, E. A. and F. E. Clark. 1996. Soil Microbiology and Biochemistry.

Academic Press: San Diego.

127

 

Payne, W. J. 1981. The status of nitric oxide and nitrous oxide as
intermediates in denitriﬁcation. Pages 85-103 in C. C. Delwiche (ed.)
Denitriﬁcation, nitriﬁcation and atmospheric nitrous oxide. Wiley-
lnterscience, New York.

Robertson, G. P. 1999. Denitriﬁcation. Pages C181-190 in M. E. Sumner et al.
(eds) Handbook of Soil Science. CRC Press: Boca Raton, FL.

Robertson, G. P. and J. M. Tiedje. 1987. Nitrous oxide sources in aerobic
soils: nitriﬁcation, denitriﬁcation, and other biological processes. Soil
Biology and Biochemistry 19: 187-193.

Robertson, G. P. and J. M. Tiedje. 1988. Deforestation alters denitriﬁcation in a
lowland tropical rain forest. Nature 336: 756-759.

Robertson, G. P., E. A. Paul, and R. R. Harwood. 2000. Nitrous oxide
production, methane oxidation, and soil carbon sequestration in
intensive agriculture: Composite effects on the atmosphere’s global
warming potential. Science (in press).

Rolston D. E., D. L. Hoffman, and D. W. Toy. 1978. Field measurement of
denitriﬁcation I. Flux of N2 and N20. Soil Science Society of America
Journal 42: 863-869.

Rolston, D. E., A. N. Sharpley, D. W. Toy, and F. E. Broadbent. 1982. Field
measurement of denitriﬁcation: lll. Rates during irrigation cycles. Soil
Science Society of America Journal 46: 289-296.

Sahrawat, K. L. and D. R. Keeney. 1986. Nitrous oxide emission from soils.

Advances in Soil Science 4: 103-148.

128

Sall, J. and A. Lehman. 1996. JMP Start Statistics. Belmont, CA: Duxbury
Press.

Sextone, A. J., T. B. Parkin, and J. M. Tiedje. 1985. Temporal response of soil
denitriﬁcation rates to rainfall and irrigation. Soil Science Society of
America Journal 49: 99-103.

Siegel, R. S., R. D. Hauck, and L. T. Kurtz. 1982. Determination of ”N; and
application to measurement of N2 evolution during denitriﬁcation. Soil
Science Society of America Journal 46: 68-74.

Skiba, U., K A. Smith, and D. Fowler. 1993. Nitriﬁcation and denitriﬁcation as
sources of nitric oxide and nitrous oxide in a sandy loam soil. Soil
Biology and Biochemistry 25: 1527-1336.

Smith, C. J. and W. H. Patrick, Jr. 1983. Nitrous oxide emission as affected by
alternate anaerobic and aerobic conditions from soil suspensions
enriched with ammonium sulfate. Soil Biology and Biochemistry 15:
693-697.

Smith, M. S. and J. M. Tiedje. 1979. Phases of denitriﬁcation following oxygen
depletion in soil. Soil Biology and Biochemistry 11: 261 -267.

Smith, M. S. and L. L. Parsons. 1985. Persistence of denitrifying enzyme
activity in dried soils. Applied and Environmental Microbiology 49: 316-
320.

Speir, T. W., H. A. Kettles, and R. D. More. 1995. Aerobic emissions of N20 and
N; from soil cores: measurement procedures using 13N-labeled N03'

and NH4’. Soil Biology and Biochemistry 27: 1289-1298.

129

Stevens, R. J., R. J. Laughlin, and J. P. Malone. 1998a. Soil pH affects the
processes reducing nitrate to nitrous oxide and di-nitrogen. Soil Biology
and Biochemistry 30: 1119-1126.

Stevens, R. J., R. J. Laughlin, and J. P. Malone. J. P. 1998b. Measuring the
mole fraction and source of nitrous oxide in the ﬁeld. Soil Biology and
Biochemistry 30: 541-543.

Stevens, R. J., R. J. Laughlin, G. J. Atkins, and S. J. Prosser. 1993. Automated
determination of nitrogen-15-Iabeled dinitrogen and nitrous oxide by
mass spectrometry. Soil Science Society of America Journal 57: 981 -
988.

Stevens, R. J., R. J. Laughlin, L. C. Burns, J. R. M. Arah, and R. C. Hood. 1997.
Measuring the contributions of nitriﬁcation and denitriﬁcation to the ﬂux of
nitrous oxide from soil. Soil Biology and Biochemistry 29: 139-151.

Tiedje, J. M. 1988. Ecology of denitriﬁcation and dissimilatory nitrate reduction
to ammonium. Pages 179-244 in A. J. B. Zehnder (ed.), Biology of
Anaerobic Microorganisms. John Wiley and Sons, New York.

Topp, E. and J. C. Germon. 1986. Acetylene metabolism and stimulation of
denitriﬁcation in an agricultural soil. Applied and Environmental
Microbiology 52: 802-806.

Vanden Heuvel, R. M., R. L. Mulvaney, and R. G. Hoeft. 1988. Evaluation of
nitrogen-15 tracer techniques for direct measurement of denitrification in
soil: ll. Simulation studies. Soil Science Society of America Journal 52:

1 322-1 326.

130

 

Vinther, F. P. 1984. Total denitriﬁcation and the ratio between N20 and N2
during the growth of spring barley. Plant and Soil 76: 227-232.

Vitousek, P. M., H. A. Mooney, J. Lubchenco, and J. M. Melillo. 1997. Human
domination of Earth’s Ecosystems. Science 277: 494-499.

Weier, K L. and J. W. Gilliam. 1986. Effect of acidity on denitriﬁcation and
nitrous oxide evolution from Atlantic Coastal Plain soils. Soil Science
Society of America Journal 50: 1202-1205.

Weier, K L., J. W. Doran, J. F. Power, and D. T. Walters. 1993. Denitriﬁcation
and the dinitrogen/nitrous oxide ratio as affected by soil water, available
carbon, and nitrate. Soil Science Society of America Journal 57: 66-72.

Wijler J. and C. C. Delwiche. 1954. Investigations on the denitrifying process
in soil. Plant and Soil 5: 155-169.

Wolf, l. and R. Russow. 2000. Different pathways of formation of N20, N2 and
NO in black earth soil. Soil Biology and Biochemistry 32: 229-239.

Yoshinari, T. and R. Knowles. 1976. Acetylene inhibition and nitrous oxide
reduction by denitrifying bacteria. Biochemical and Biophysical

Research Communications 69: 705-710.

131

lrill)jiiliiiiilil