GLOBAL BIOGEOCHEMICAL CYCLES, VOL. 7, NO. 2, PAGES 275-290, JUNE 1993
COMPARISON
OF CARBON
DYNAMICS
IN
TROPICAL
AND TEMPERATE
SOILS USING
RADIOCARBON
MEASUREMENTS
SusanE. Trumbore1
Center for Accelerator Mass Spectrometry
LawrenceLivermoreNationalLaboratory
Livermore, California
Abstract. The magnitude and timing of the
responseof the soil carbonreservoirto changesin
landuseor climateis a largesourceof uncertaintyin
global carbon cycle models. One method of
assessing
soil carbondynamics,basedon modeling
theobserved
increaseof •4Cin organicmatterpools
during the 30 years since atmosphericweapons
testingended,is described
in thispaper.Differences
in the inventory and residencetime of carbonare
observedin organicmatter from soilsrepresenting
tropical (Amazon Basin, Brazil) and temperate
(western slope of the Sierra Nevada mountains,
California) forestecosystems.The majorityof the
organiccarbonin theupper22 cm of thetropicalsoil
(7.1 kgC m-2) hasresidencetimesof 10 yearsor
less, with a minor componentof very refractory
carbon.
The estimated annual flux of carbon into
andout of the soil organicmatterin thishorizonof
themineralsoil,basedon modelingof the14Cdata,
is between1.9 and 5.5 kgC m-2 yr-•. In contrast,
organicmatterin the temperatesoil over a similar
depthinterval(0-23cm;5.2kgCm-2),is madeupof
approximately equal amounts of carbon with
residencetimes of 10, 100, and 1000 years. The
•Now at Department
of Geosciences,
University
of California, Irvine, CA 92717-3100.
Copyright1993
by theAmericanGeophysical
Union.
estimated annual flux of carbon into and out of this
soilis 0.22 to 0.45kgC m-2yr-•. Rapidturnoverof
organicmatterwith density<1.6 - 2.0 g crn-3
contributes
a majorcomponent
of theannualflux of
carboninto and out of both soil types. Hydrolysis
of mineralsoil organicmatterof density> 1.6-2.0 g
cm-3 removed•4C-enrichedcomponentsfrom the
temperate
soilbuthadnoeffectonthe14Ccontent
of
the residuein 0 - 22 cm layer of the tropicalsoil.
The resultspresentedhere showthat carboncycle
modelswhich treatsoil carbondynamicsas a single
reservoir with a turnover rate based on radiocarbon
measurements of bulk soil organic matter
underestimatethe annualfluxes of organicmatter
throughthe soil organicmatterpool, particularlyin
tropicalregions.
INTRODUCTION
Role of Soilsin the Global CarbonBudget
Recentdebateover the importanceof soil organic
matterin the globalcarbonbudget[Tanset al., 1990;
Schlesinger, 1990, Prentice and Fung, 1990;
Jenkinsonet al., 1991] has emphasizeda lack of
fundamentalknowledge of soil carbon dynamics.
An understandingof the degreeand timing of the
responseof this carbon reservoir to perturbations
associated
with climaticor land usechangerequires
knowledgeof both the inventoryof carbonin soils
and the turnover rate of carbon in soil organic
matter. An estimated 1300 to 1500 GtC (1GtC =
1025gC) is storedgloballyasorganicmatterin the
Papernumber93GB00468.
0886-6236/93/93GB-00468
$10.00
upper meter of mineral soils [Schlesinger, 1977;
Post et al., 1982]. This is roughly twice the
276
Trumbore:CarbonDynamicsin TropicalandTemperateSoils
estimatedstorageof carboneitherin the atmosphere
or in the living biosphere(both approximately700
GtC [Atjay et al, 1979]), and 300 times the annual
fossilfuel input[Intergovernmental
Panelon Climate
Change, 1990]. Little informationis availablefrom
which to determine the residence time of carbon in
the soil organicmatterreservoir,especiallyregarding
its variation with soil forming factors such as
climate,topography,time, parentmaterial,andland
use. In large part, the lack of knowledgeis due to
the complex, heterogeneousnature of soil organic
matter (see reviews in the works by Stevenson
[1982] and Stevenson and Elliott [1990]). Soil
organicmatterincludesa wide spectrumof organic
compounds, from labile components, such as
relativelyfreshvascularplantmaterialandmicrobial
biomass, to refractory components which
accumulateslowly over thousandsof years.
In order to understand the role of the soil carbon
pool in the global carboncycle, it is necessaryto
quantify the accumulation, transformation,
translocation,and eventual decompositionof soil
organicmatteron time scalesrangingfrom seasonal
to millennial.
Models of soil carbon turnover at the
ecosystem level [Jenkinson and Raynor, 1977;
Parton et al., 1987; Emanuel et al., 1985; Van
Bremen and Feijtel, 1990; Jenkinsonet al., 1991],
differentiate soil organic carbon into fast (annual
turnover), slow (decadal to centennial turnover) and
passive(millennial and longer turnover)pools. As
thesemodelshavebeensuccessfulin explainingsoil
carbondynamicsin the ecosystemsfor which they
have been developed, a method with which to
determine the distribution of soil organic carbon
among thesepools in a variety of ecosystemsand
soiltypesis greatlyneeded.
Quantification
of SoilCarbonPoolsUsing14C
Carbon 14, the radioactiveisotopeof carbon,is a
usefultool for decipheringthe dynamicsof carbon
cyclingin soils[Stoutet al., 1981; Balesdent,1987;
Scharpenseelet al., 1968a, b, 1989; Trumbore et
al., 1989]. New carbon is added to soils as vascular
plantmaterial,with a 14C/12Cratiocloseto thatof
contemporary(that year's) atmosphericCO2. The
degreeto which the 14C/12Cratio in soil organic
matter differs from the vascularplant matterfrom
whichit is derivedreflectsthemeanageof carbonin
soils, often used to calculate a Mean Residence Time
(MRT) [Paul et al., 1964]. Severalproblemsexist,
however,with usingonly the 14C-derived
MRT of
bulk soil organic matter to interpret soil carbon
dynamics. First, the presence of unknown
quantitiesof bomb 14Cin soilssampledsincethe
mid-1960s will cause the MRT to be underestimated
in modern soils.
Even where archived soils
(sampledprior to 1960) are available,variationsin
the prebombatmospheric•4C content,combined
with the analytic-al uncertainty of the •4C
measurement, limit the use-fullness of radiocarbon
to study carbon pools with MRT less than several
hundredyears. Second,the •4C MRT cannotbe
usedto directlydeterminethe totalflux of CO2 to the
atmosphere from soil organic matter. The
calculation
of CO2fluxesdirectlyfromthe•4CMRT
ignoresotherlossmechanisms,
suchas the transport
of dissolvedor particulateorganicmatterout of the
soil profile. Theseprocesses
may be the ultimate
factordefininglong-termresidencetime of carbonin
refractoryorganicmattercomponents[Van Bremen
and Feijtel, 1990]. Finally, the MRT basedon the
14Cageis a misleading
measure
of theannualfluxes
of carboninto and out of soil organicmatter. This
resultsprimarilyfrom thefact thatthe averageageof
a carbon atom in soil organic matter, a reservoir
madeup of a suiteof components
with vel5, different
residencetimes, does not necessarilyequal the
averagetime spentby a carbonatomin soil organic
matter (the turnover time [Balesdent, 1987]). For
example,a •4C ageof 1000yearsobtainedfor bulk
soil organiccarboncannotdistinguishbetweenthe
homogeneous
case,in which all of the carbonin the
soil has a turnover time of 1000 years, and the
inhomogeneous
case,in which 75% of the organic
carbonhas a turnovertime of 20 years,and 25% a
turnovertime of 5000 years. Although thesetwo
soils would have the same MRT based on bulk •4C
content, the annual flux of carbon into and out of the
soil, and the response of soil carbon to a
perturbation,will differ greatlybetweenthesetwo
cases.
Two approaches
havebeendeveloped
to use•4C
data to break soil organicmatter down into more
labile and more refractoryconstituents.The first
methoduses•4Cmeasurements
of operationally
defined fractionsto partitionsoil organicmatter into
constituentlabile and refractorypools [Paul et al.,
1964; Campbellet al., 1967;Martel andPaul, 1974;
Goh et al., !976, 1977, !984; Scharpenseel
et al.,
1968a, b, 1984; Trumbore et al., 1989, 1990]. The
second
method
is
based
on
the
observed
incorporationof bomb-producedradiocarboninto
soil organicmatterduringthe 30 yearssincethe end
of atmospheric
weaponstesting[AndersonandPaul,
1984; O'Brien and Stout, 1978; O'Brien, 1984,
1986; Harkness et al., 1986; Trumbore et al., 1989,
1990; Jenkinsonet al., 1992]. This approach
requires14Cmeasurements
of soil organicmatter
sampled prior to 1960 for comparison with
contemporarysamples. A large increasein •4C
content over the past 30 years indicates that
significantportionsof the soil organicmatter are
exchangingcarbonwith atmospheric
CO2 on decadal
and shortertimescales. A comparisonof •4C in
prebomband postbombsoilsalsoprovidesthe best
meansto testthe usefulness
of variousoperationally
definedseparation
procedures.
Measurement of •4C by accelerator mass
spectrometry (AMS) has greatly reduced the
Trumbore:CarbonDynamicsin TropicalandTemperateSoils
difficultiespreviouslyinvolvedin studiesof 14Cin
soil organic matter. Archived soil samples are
generallyso small that they do not containenough
carbonto measure•4Cby decaycountingmethods.
Using AMS, measurementof •4C in several
fractionsof soil organicmattermay be madeusing
only several hundred milligrams of soil. An
additional advantageof AMS is the high rate of
samplethroughput.Up to 100 samplesmay now be
measuredin a 24-hourperiodusingAMS [Davis et
al., 1990].
277
madein thisstudyfor environments
similarto those
studied by Jenny: moist tropical forest (Amazon
Basin, Brazil) and dry temperateforest (western
slope of the Sierra Nevada mountainsin central
California). Information on the locations, climate
and vegetation characteristicsof these sites are
summarizedin Table 1. More specificinformation
on soil chemical and physical properties are
presented
togetherwith •4Cdatain Table2.
The same number of measurements
TemperateSoil California
would require several monthsto a year in many
countinglaboratories.
Samplesof a soil namedthe Musick SandyLoam,
described and collected in 1959, were obtained from
SAMPLES
the archives at the U.S. National Soil Survey
InvestigationsLaboratory. This soil is classifiedas
an Ultisol (Xerult), signifying a soil developedin
Mediterranean climate, with an argillic horizon
(moderateclay contents),low base saturationand
red to yellow color signifyingthe presenceof iron
and aluminum hydroxides [Brady, 1984]. A
detailed profile descriptionmay be found in Soil
SurveyInvestigationsReport 24 (California). The
samplingsitewaslocatedon the westernslopeof the
Sierra Nevada mountains, Amador County,
California,at an elevationof approximately1300m.
The vegetationat thiselevationconsistsprimarilyof
oak and pine, with grassesand shrubs(seeTable I).
A contemporarysoil profile was collectedin 1990,
approximately2 m from the locationof the original
soil pit (as located by the land owner). Some
disturbanceof the site vegetation had occurred
A comparisonof the organiccarbondynamicsin
litter and soils from the western Sierra Nevada and
tropical regions (Costa Rica and Columbia) was
publishedby the late Hans Jenny [Jenny et al.,
1949; Jenny, 1950]. He observedthat, although
overall input and decompositionratesof freshplant
material are higher in the warm, humid tropics,
tropical soils contain more carbon than the Sierra
soils. The ratio of carbonin detritallayersto that in
the mineralsoil is muchsmallerin tropicalsystems
thanin temperatesystems[Jenny,1950], indicating
large differencesin the way carboncyclesin these
soil systems.Given the largeprobabledifferencein
soil carbondynamicsbetweenthesetwo soil types,
comparisonsof the radiocarboncontentof organic
matter in prebomb and postbombsoil pairs were
TABLE 1. Summaryof generalcharacteristics
for thetropicalandtemperatesoilsusedin thisstudy.
Soil/
Climate
Parent
Vegetation
Temp.,Precip. material
Life Zone
NPP
present kgCm-2yr-1
Oxisol (Kaolinitic Yellow Latosol) 22 to 26øC
Moist TropicalForest
>200cmyr-1
sediments*
Ultisol (Musick)
Dry TemperateForest/Scrub
granodiorite ponderosa
pine, 0.3 - 0.5
SoilID
Location
-5 to 20øC
terrafirme
Soil C
kgCm-3
1.0
11.4
tropical
forest
20- 30cmyr-1
Year Sampled
7.1
liveoak,bluewildrye
Relevent
References
Oxisol 1
Curua-Una, Brazil
1959
303; profile 24 [Sombroeck,1966,p. 129]
Oxisol 2
Belem, Brazil
1959
300; profile 25 [Sombroeck,1966, p. 130]
Oxisol 3
Manaus, Brazil
1978; 1986
Leenheer, 1980; Trumbore et al., 1990
Ultisol 1, 2
AmadorCounty,
1959, 1992
SSIR No. 24 (S59-3-13) [Jennyet al., 1949]
More specificinformationandsoil decsriptions
may be foundin thereferences.Data on NPP andsoil carbon
storagefor theseecosystem
typesaretakenfrom the literature[Atjay et al, 1985;Postet. al, 1982].
* thereis somedebateasto to whethersoilsin theAmazonBasinaredevelopedon fluival sediments
(Sombroeck,1966)or represent
very thickprimaryweatheringsequences
(Irion, 1984).
Trumbore:
CarbonDynamics
in TropicalandTemperate
Soils
278
CX• tt'•
i
•++
'
''7,'7,
++
''7,
AAAA
o•5•
i
i
i
i
i
i
i
i
Trumbore:
CarbonDynamics
in TropicalandTemperate
Soils
279
duringthe 30 yearsbetweensamplecollections.In
the 1950s the understoryin this area was burned
regularly to eliminate underbrush.Burning was
tropicalsiteswere undisturbedprimaryforestat the
time of sampling.
discontinued in the late 1950s, and manzanita scrub
METHODS
grew up in the area. The scrubwas cut backin the
late 1980s.
TropicalForest Soils,AmazonBasin,Brazil
Samplesfrom two tropical forest soil profiles,
collected in Brazil in 1959, were obtained from W.
Sombroeck (then Director, International Soil
Resourceand InformationCenter,Wageningen,the
Netherlands).The profilesfrom which the samples
were takenare describedby Sombroeck[1966, profiles 303 and 300].
Both soils are described as
Oxisols(Orthod), signifyinga wet, warm, climate,
and a high degree of weathering [Brady, 1984].
Thesesoilshave high clay contents(predominantly
All archivedsoils were storeddry, and had been
sieved to remove components> 2 mm, including
large root fragments. Contemporarysoil samples
collected for this study were sieved to remove
components>2 mm, and eitheroven-driedor stored
in a refrigeratorprior to analysis.
Fractionationof the soil organicmatterfollowedthe
methoddescribedby Trumboreet al. [1989] and is
illustratedin Figure 1. Initial acidificationto remove
calcium carbonate(not presentin any of the four
profiles) was followed by a density separation
[Sollins et al., 1983] to quantitatively remove
vascular plant debris, charcoal and easily soluble
organic matter. A zinc bromide solution with
kaolinite), mixed with abundantiron and aluminum
densityof 1.6 g cm-3wasusedfor the tropicalsoil.
hydroxides. The first of the archivedtropicalsoil
samples(Sombroeckprofile 303) were collectedin
The temperatesoil, collectedand analyzed2 years
after the tropicalsoil, was fractionatedusinga new
product,sodiumpolytungstate(densityof 2.0 g cm-
Curua Una, Brazil, near the center of the Amazon
Basin. Two samples from this profile were
available,integrating0-22 cm and22-60 cm depths.
The secondarchived Oxisol profile (Sombroeck
3). Sodiumpolytungstate
is preferablebecause
it is
number300) was collectednear Belem, Brazil, close
Both heavy liquids solubilizesome(small) portion
of the soil organicmatter,which is not quantifiedin
the procedure.Three to six gramsof driedsoil were
shakenvigorouslywith the densesolutionin a 25mL centrifugetube. The mixture was centrifuged,
to the mouth of the Amazon River. Only the 0-32
cm layer from this soil was available.It has much
lower clay contentsthan the Curua Una soil (Table
2).
The contemporary
tropicalsoilprofilewascollected
inert (not poisonousor corrosive),and may be used
to attainliquidsof densitygreaterthan2.0 g cm-3.
in 1986 from the Reserva Ducke, about 17 km
northeast of Manaus, Brazil, and the site of the
Bulk Soil
ABLE IIa andIIb experiments[Hardsset al., 1990].
In addition,two samplesof humic matter,extracted
Density
Separation
using
heavyliquid
from a soil in the Reserva Ducke in 1978, were
obtainedfrom J. Leenheer(U.S. GeologicalSurvey;
describedby Leenheer[1980]). Soilsin theReserva
Ducke sampledfor this studywere alsodescribedas
Oxisols,with clay contentssimilarto the CuruaUna
soil.
All threeof the tropicalsoilsobtainedfor thisstudy
are thoughtto be developedon fluvial clay deposits
of Plioceneor Pleistoceneage [Sombroeck,1966],
althoughit has also been suggested[Irion, 1984]
that the soilsfrom this regionmay representin situ
weatheringprofiles. Vegetationat all sitesis terra
firme (not seasonally flooded) tropical forest.
LightFraction
undecayed
vascular
plantmaterial,
charcoal
study cannot be considereddirectly comparable.
Sombroeck'sprofile 303 (fi'om Curua Una) will be
usedto provideonelimit with which to comparethe
and
1986
Reserva
Ducke
soils.
cell wall debris,
mineral bound
organic
matter
6N HC1
95 øC
18 hours
Nonetheless, the three Oxisols obtained for this
1978
Dense Fraction
The
justification for this comparison is seen in the
comparablecarbon and clay contentsof the two
hydmysis
residue
soils,andin thesimilarityin pre-bomb•4Ccontents
measuredin the two amhived Oxisols (see Table 2),
despitetheirdifferencesin locationandparticlesize.
As far as it was possibleto ascertain,all of the
Fig. 1. Fractionation
schemeusedfor soilorganic
matter, as described in the text.
280
Trumbore:CarbonDynamicsin TropicalandTemperateSoils
and the supernatefiltered to removefloatinglowdensitymaterial.The filteredsolutionwasreusedto
exhaustively
extractremaininglow densitymaterial
(3-6 resuspension-centrifugationcycles were
requiredin near-surface
layers).The denserresidue,
includingorganicmatter stabilizedby sorptionto
mineral
surfaces and microbial
cell wall debris
[Spycheret al., 1983],wasrinsedthreetimeswith
distilled water by resuspension/centrifugation,
vacuumdried, and ground.This fraction,hereafter
referred to as the dense fraction, was the substrate
for further fractionation.
Acid extractioninvolvedhydrolysisof the soil in
6N HC1 for 18 hours at 95øC. The soil was then
centrifuged,and the hydrolyzateseparated. The
residue was rinsed three times with distilled water,
then vacuum dried.
The tropicalsoil samplesfrom Reserva Ducke
obtainedby Leenheer[1980] representthe organic
matterextractedby base(NaOH), andsubsequently
purified using XAD-8 resin [Leenheer, 1980].
Comparisonsof base and acid extractedorganic
carbonfrom theupper20 cm layersof the prebomb
tropical soil samples [Trumbore et al., 1990]
showed no difference
in 14C content
between
unextracted, base extracted and acid extracted soil in
upper mineral soil layers (0-22 or 0-32 cm depth
interval). With increasingdepthin the mineralsoil,
the base extracted material contains increasingly
more 14C than the residue [Trumbore et al., 1990;
alsounpublished
data, 1992]. Thereforethe 14C
contentof the extractsobtainedfrom Leenheermay
be greater than or equal to the unextracteddense
fractionorganicmatter to which they are compared
in this paper.
Samplesfor 14Canalysiswerecombusted
(900øC
for one hour) with CuO wire in sealed, evacuated
quartz tubes[Buchananand Corcoran, 1959]. The
evolved CO2 was cryogenically purified and
manometricallymeasuredto determinethe weight
%C in each fraction. Comparison of %C data
obtained by sealed tube combustion with those
obtainedusing a commercialCN analyzershowed
good agreement between the two methods. For
someof the archivedprofiles,previouslydetermined
%C data were available (see referencesin Table 1).
Althoughagreementbetween%C valuesobtainedin
this studywith thesepreviousresultswas generally
good, some discrepanciesexist. The most likely
cause for the differences
lie in the use of different
methodsfor determining%C [Jenny, 1950]. The
%C was convertedto a carboninventoryusingbulk
densitydata obtainedduring profile collection,or
estimatedusing equationsby Zinke et al [1984].
Errorsdueto the problemswith discrepancies
in %C
andestimationof bulk densityare largeanddifficult
to estimate,but, in general,wheresoil bulk density
is known and %C determinations agree with
previousdeterminations,
areestimatedat 10-20%of
thetotalreportedsoilcarboncontent.Table2 shows
the values of %C and bulk density used for the
calculationof soil carbondensity.
CO2 evolvedfrom high-temperature
sealedtube
combustionwas catalyticallyreducedto graphite
AMS targetsusing the methodof Vogel et al.
[1987]. Carbon14 AMS measurements
were made
at thePSI/ETHfacilityin Ztirich,Switzerland
[Suter
et al, 1984], and at the Centerfor AcceleratorMass
Spectrometryat LawrenceLivermoreNational
Laboratory
[Daviset al., 1990].813Ccorrections
to
14Cdatawere madeusingmeasured13C/12Cin
Ztirich; for CAMS determinations
•513Cwere
assumedto be-25%o. The error in 14C is less than
1.5% of Modern for all samples. The sourceof
errorfromisotopicmeasurements
is alwayssmallin
comparison
withtheerrorinvolved
in calculating
the
carboninventoryin the soil. Carbon 14 data are
expressedas A14C , the deviation,in parts per
thousandof the 14C/12Cratio in the samplefrom
that of an absolute standard (oxalic acid decay
correctedto 1950 [StuiverandPolach,1977]):
14C,
A14C
__12
Csam1'1000
1-••'
c
(1)
12 Cstd
Measurementof the activity of 137Csin soil
samples
weremadeby gammacountingof bulksoil
samples. Cesium 137, an isotopeproducedby
atmospheric
weaponstesting,hasa half life of 30
years. In soils,it is primarilyassociated
with clay
minerals and organic matter [Monaghan, 1984;
Graustein and Turekian, 1986; D6rr and Mtinnich,
1989]. Comparisonof 137Cswith the vertical
distribution of bomb-produced 14C yields
informationabouttheprocesses
transporting
carbon
vertically in the soil profile [D6rr and Mtinnich,
1989; Trumbore et al., 1989].
RESULTS
DensityFractionation
of SoilOrganicMatter
Figure2 (a throughf) plotsthecarbondensityand
14Ccontentof organicmatterin the temperatesoil
(SierraNevada Ultisol) with depth. The resultsare
separated
intocomponents
of differentdensity:<1.0
g cm-3, 1.0 to 2.0 g cm-3, and>2.0 g cm-3. The
low-densityfractions(<2.0 g cm-3) constitute
roughly50% of thetotalsoilcarbonin the0-23 cm
layer(Figures
2aand2b;Table2). The14Ccontent
of thenearlyunaltered
vascularplantmaterialwhich
passes
througha 2-mmsieve,(<1.0g/cm3;Figure
2d) hasA14Ccloseto atmospheric
valuesat thetime
of samplingfor the 19590 - 23 cm layer. The 1990
value of +270 %ofor the <1.0 g cm-3 fractionis
greaterthan1990atmospheric
14CO2values(+ 172
Trumbore:
CarbonDynamics
in TropicalandTemperate
Soils
C content,
gCm'2cm'1
0
100
200
300
400
281
C content,
gCm'2cm'•
$00
0
100
200
300
400
$00
•S?Cs
.oo
'
0.00
g-•3.00
0
203t
•
20..
• 40
40 )
40
60
so
-
60• )ensi
so Bhi • •
6O
_
-
C. 137Cs
J
8O
I
i
i
i
AI4c %0
•00
-200
0
200
-400
-200
0
200
-400
-200
200
o
,,t
øt
_
• 40
4O
60
60
-
_
D. Density < 1.0
8O
i
I
I
I
I
ß Density 1.0I
I
I
!
!
2.0
I
I
F. Density > 2.0
8O
I
I
I
I
I
Fig.2. Results
of densityfractionation
for theUltisol,SierraNevada,California.Carboninventory
(in gCm-2per
cmdepth)
and•4C(A•4C,%0)content
areplotted
versus
depth(incentimeters).
Thepointsrepresent
themidpoint
of theintervalrepresented
by eachsample;the "errorbars"definethe verticalextentof eachlayer. Archived
samples(1959) are shownas filled circles;contemporary
samples(1990) as opencircles. The figuresshow
carboncontained
in thefractionwith density(a) lessthanand(b) greaterthan2.0 g cm-3, and(c) 137Cs
contentof
thebulksoil. Carbon14 contents
of threedensityfractionsare shown:density(d) < 1.0 g cm-3;(e) between1.0
and2.0 g cm-3;and(f) >2.0 g cm-3(the"dense"
fraction).
_+10%0[Aravenaet al., 1992]). Thesedatamaybe
of the site indicates that it was last burned in the late
explained if the combinedresidencetime of lowdensitycarbonin plantsandin the detritallayersis
about 20 years (see Figure 3, and discussionin
1950sor early 1960s,whenthis would havebeena
masonable
valuefor atmospheric
or plant•4C. If we
modelingsection). Increases
in •4C in the <1.0 g
+ 107 %0)represents
a mixtureof remnantcharcoal
(A•4Cof +6%0)andplantmattersimilarto thatin the
<1 g cm-3fraction(A•4Cof +260%o),thencharcoal
makesup roughly40% of the carbonin thisfraction.
The bulkof thesoilorganicmatterin thetemperate
ecosystem
mineralsoil is in the dense(p >2.0 g cm3) fraction(Figure 2b). Carbon 14 data for this
fraction (Figure2f) showthat the organiccarbon
½m-3 fraction between 1959 and 1990 are less
pronounced
in the deepersoil horizons,indicating
slowerdecomposition
of plantmaterialdeeperin the
soil. Organicmatterwith densitybetween1 and2 g
½m-3 (Figure2e) consistsof a mixtureof charcoal
andaltered,butstill recognizable,
plantmatter. •4C
contents
in thepre-bomb
soilprofileareslightlyless
than those in the <1.0 g cm-3 fraction, and the
increase in •4C between 1959 and 1990 is smaller.
Charcoal
pickedoutof the 19901-2 g cm-3density
fractions in the 0-23 cm interval had A•4C of 6 +_ 10
%0.Anecdotal
information
fromtheproperty
owner
assume
thetotal19901-2 g cm-3fraction(A•4C=
associated with the dense fraction contains carbon of
greateragethanlow-densityfractions.The A•4Cof
densematerialin the 0-23 cm layer increasesfrom
-53%0in 1959 to +70%0in 1990, a smaller increase
thanwasobservedin thelow-densityfractions.
282
Trumbore:CarbonDynamicsin TropicalandTemperateSoils
A plot of •37Csactivity(measured
in thebulksoil;
data from R. F. Anderson,Lamont-DohertyGeologicalObservatory,1991) is shownfor comparison
withthe•4Cdata(Figure2c). Measurable
quantities
of •37Csarepresentto only23 cm depthin thesoil,
while bomb •4C is observeddeeperthan 23 cm,
particularlyin low-densityfractions. This indicates
that the sourceof low-densitymaterial deeperthan
23 cm in the soil is from root turnover (or soluble
transport) rather than the downward mixing of
surface detrital material.
The data in Table 2 show substantial differences
the distribution
of carbon and •4C between
in
the
tropical and temperatesoils. While carboninventories in the upper 22-23 cm are similar for both
soils,the tropicalsoil storesmore carbonin deeper
layers. Low-densitymaterial (for the tropicalsoil,
<1.6 g cm-3) makesup a only 15% of the totalsoil
organic matter in the 0-22 cm layer (comparedto
50% for a similar depth interval in the temperate
soil). The Reserva Ducke soil (1986) low-density
fractionhasA•4C of + 185%o,closeto the expected
1986 atmosphericvalue (about+ 190%o[Manninget
al., 1989]), and the carbonin surfacedetritallayers
(+210%o- 220%0). This indicatesthat turnoverrates
in the tropical soil may be close to annualin lowdensityfractions. Densefractionsin the tropicalsoil
containedless •4C in the prebombsamplesthan
equivalentdepthsfor the temperatesoil.The increase
in •4C between1959 and 1986, especiallyin the 022 cm layer, was greaterin the tropicalsoil thanin
the temperate soil. Thus a larger portion of the
tropicalsoil organiccarbonmustbe in poolswhich
FractionationbyAcid Hydrolysis
Acid hydrolysisof mineralsoil organicmatterhas
been shown by several investigatorsto leave a
residuedepletedin •4C [Scharpenseel
et al., 1968a;
Campbell et al., 1967; Goh et al., 1977; Anderson
and Paul, 1984; Trumbore et al., 1989, 1990].
Extractionof thep >2.0 g cm-3fractionwith strong
acid hydrolyzesproteins,amino sugars,and some
carbohydrates[Jawsonand Elliott, 1986]. As these
are presumably compounds readily utilized by
microbes, acid hydrolysis may be expected to
remove more labile, rapidly cycling carbon into
solution,leaving a more refractory,slowercycling
residue [Stout et al., 1981, and referencestherein].
The acid hydrolysistreatment,however,may also
causefunctionalgroupson high molecularweight
polymersto be removedor exchanged,thusperhaps
addingmore •4C rich material to otherwisemore
refractory carbon, or solubilizing refractory
components. To test the efficiency of acid
hydrolysis in isolating a "passive"soil organic
matterpool,thenonhydrolyzable
components
of the
temperateand tropicalprebombandpostbombsoils
were compared.
The nonhydrolyzablecarbonin the temperatesoil
contained 15% to 30% of the total soil cm'bon at each
depthin the soil profile (Table 2). The residueafter
hydrolysisin HC1 containedless radiocarbonthan
theunhydrolyzeddensefractionat all depthsin both
the 1959 and 1990 temperatesoils. Nonhydrolyzablecarbonmadeup a largerportion(40-80%) of
the tropical soil organic carbon. There were,
however,no significantdifferencesin the A•4C of
turn over on decadal or shorter time scales. Cesium
the total dense fraction and the residue after 6N HC1
137 was only detectablein the detritallayersof the
hydrolysisin the 0-22 cm layersof eitherthe 1959
Ducke soil (J. Beer, EAWAG,
Switzerland,
personalcommunication,1990). However, fallout
of •37Cswasvery low in the southern
hemisphere,
so it is not a reliableindicatorof verticaltransportin
this soil.
or 1986 Oxisol.
MODELING
The Mean ResidenceTime (MRT) of soil carbonis
calculatedusing a decayconstantderivedfrom the
Fig. 3. Schematic
depictionof thefour-component
modelusedto reproduce
the 14Ccontents
in prebomband
postbomb
soils.The plotsarerepresentations
of A•4Cversustime(1950-2000A.D.) The atmospheric
•4CO2
curve (depictedhere for the Southernhemisphere[Manninget al., 1990]) is at the far left; annualadditionsof
carbonto eachof thefoursoilcomponents
(depicted
in thefourverticallystacked
boxes)arelabeledwiththe14C
contentof the atmospherefor that year (northernhemispherevalues,usedfor the Sierra Nevada soil are taken
from Levin et al. [1985] and Tans [1981]. Each of the four componentsis assumedto be homogenous(i.e.,
represented
by a singleturnovertime for carbon,given as z in Figure3) and at steadystate. The turnovertime
fixesthe 195514Ccontentfor eachof thefourfractions,
aswell astheincrease
in •4Cduring1955- 2000A.D.
Thffq4Ccontentof themodeledsoilorganicmatter,thesumof thefourcomponents
weightedby theamountof
carbonin eachpool (numberin theupperfight-handcomerof eachsoil componentbox), appearsin thebox at the
far right.
To run the model, the amountsof carbonin eachof the four components
is adjusteduntil the modelreproduces
boththe observedprebombandpost-bomb•4C contentsfor the totaldensefractionfor a givensoil anddepth
interval. The annualfluxesof carbon(numbersover the arrows)are the productof the amountof carbonin each
componentand 1/z As the soil is assumedat steadystate,annualinputs= outputs.
Trambore:CarbonDynamicsin TropicalandTemperateSoils
I
I
I
I
I
I
I
i
I
I
I
i
i
I
I
I
I
283
284
Trumbore:CarbonDynamicsin TropicalandTemperateSoils
14Ccontentof prebomborganicmatterat steady
state:
R=
rd
(r d + r 14)
, whereR = A14C/1000
+ 1,
and rd and r14c are the decay constants for
decomposition
(first order)and 14Cradioisotope
decay[Trumboreet al., 1992]. This relationmay be
solvedfor rd when R is measured;the MRT for soil
organic matter equals 1/rd [Paul et al., 1964].
pools was unsuccessful,a four-componenttimedependentbox model was used to simulate the
increase in 14C between 1959 and 1990 for the
densefractionsof the two soils. Figure 3 illustrates
the operationof the model. The soil carbon is
arbitrarilyassumedto consistof four components
with MRT's of 1 or 10, 100, 1000, and 10,000
years. In reality,themake-upof soil organicmatter
MRT's calculated from the 14C contents of the 1959
maybe betteor
represented
as a continuum,
asit is
dense fractions for the temperate and tropical
prebombsoilsare 470 and990 years,respectively.
The annualinputs(I) requiredto sustainthe carbon
inventoriesobservedin theselayers, assumingthe
carbon is homogeneous,with rd equal to the
turnovertime, would be 0.005 kgC m-2 yr-1 and
0.006 kgC m-2 yr-• respectively,for the temperate
(2.6 kgC in 0-23 cm densefraction) and tropical
(5.7 kgC in 0-22 cm) soils.
The inadequacyof this approachis shown when
comparingthe observedincreasein 14C between
1959 and 1986/90 to thatpredictedusingthe values
for I andrdderivedfromtheprebomb
14Ccontent
of
the total densefraction. In order to reproducethe
increases
observedin 14Cthroughthe 1959-1990
period,a time-dependentbox model (see Figure 3)
was usedto track the annualadditionalof new plant
matter to the soil (I) and its lossthroughdecay(at
rate rd). The carbon added to the soil annually is
assumedto have a 14C/12C ratio equal to the
atmospheric14CO2contentfor thatyear [Levin et
al., 1989; Manning et al., 1989]. Using this model
and assumingthe turnoverof soil organiccarbonis
homogeneous
with a decayconstantequalto 1/470
years,thepredictedincreasein 14Cfor thetemperate
modeledby AgrenandBosatta[ 1987]andBalesdent
soil dense fraction
is
-53%o in 1959 to -51%o in
1990. For the Oxisol (rd - 1/990 years), the
predictedincrease
is from-110%oto -100%•. The
increases observed in Table 2 are much larger,
supporting
the conceptof organicmatterin soilsas
a mixture of componentswith different turnover
rates.
The effectiveness of the acid hydrolysis
fractionation
methodmaybe assessed
usinga similar
comparison
of prebombandpostbomb
radiocarbon
data. For example,thepredicted
increase
in 14Cfor
the 6N hydrolysisresiduefor the 0-22 cm layer of
thetemperate
soil (Ultisol)shouldbe from-108%•
(1/rd about 1000 yr) to-100 %o.The observed
increasewas greater,to -5%o. Hydrolysisof the
tropicalsoil densefractionin 6N HC1producedno
change in the radiocarboncontent of the nonhydrolyzedorganicmatter. Thus, the residual
organicmatterafteracidhydrolysis
stillrapresents
a
mixtureof more rapidly and moreslowly cycling
material. Althoughnotin itselfsufficientto predict
the carbondynamics,the 6N HC1 hydrolysisdoes
isolate a more 'passive'fraction in the temperate
soil.
As a chemicallybasedfractionationschemefor
separating
organicmatterintolabileandrefractory
[1987]. However, the simulation as used here is
suitable for the task of emphasizingdifferences
betweenthe carbondynamicsin the two soils. The
amountsof carbonin eachbox were adjusteduntil
the model output reproduced(1) the observed
inventoryof carbonin the total densefraction,(2)
theprebomb
(steady
state)14Ccontent
of thedense
fraction,and(3) theobserved
increase
in 14Cin the
dense fraction between prebomb and postbomb
samples. A family of solutionsexistswhich will
reproducethe observations
in eachcase(i.e., a range
of allowable amounts of each age fraction may
providethesame14Ccontent
in 1959and1986or
1990). Severalof thesesolutionsare plottedfor the
0-22 cm layers of the Ultisol (Figure 4a) and the
Oxisol(Figure4b). Figures4c and4d plottherange
of allowed values for each age component.The
rangesshownin Figures4c and4d maybe reduced;
if the size of one of the componentsis fixed, the
allowed ranges of the remaining three will be
smaller.
Figure4 showsresultsonlyfor thedensefraction
of the 0-22 or 0-23 cm layersof the soils. For the
tropicalsoil (Figure4b), the0-22 cm layerwassplit
into two levels. The 1977 and 1986 A14C in the 0-
10 cm layer organic matter fall directly on the
(southern
hemisphere)
atmospheric
14CO2curve,
indicatingthat the residencetime of carbonin this
soil layer could be 1 year or less. That result,
however,is not consistentwith a 1959 A14Cvalue
-110%•for thetropicalsoil. If theprebomb
•4Cdata
from the Curua Una 1959 site may be compared
directly with the ReservaDucke 1986 data, some
componentof very refractorycarbonmustbe present
in thedensefraction.The amountby whichthe14C
content of the soil is diluted by that refractory
componentmust be compensatedfor by having
someportionof the soil carbonturn over on 10 - 20
yeartime scales(i.e., a component
whichhas14C
valuesin 1986 which are greaterthancontemporary
atmospheric
values). Modelswhichhavesignificant
portionsof the soil organic matter in the 100 and
1000 year pools cannotreproducethe observations
for the Oxisol.
In contrast,the densefractionof the temperatesoil
is bestmodeledusingapproximatelyequalportions
of the 10, 100, and 1000 year pools.No difference
is observed between 0-12 and 12-23 cm layers,
which may be due to vertical mixing of the soil
(evidenced
by 137Csdatain Figure2f). In orderto
Trumbore:CarbonDynamicsin TropicalandTemperateSoils
1,000
I
I
atmos- •
800
I
1,000
I
4A. Temperate Soil,
pher
285
I
I
I
I
4B. Tropical
800
0-22
Soil,
cm
atrnos-
600
600
400
400
200
200
• /•.•f.:'•" 10-22
-200
I
1950
1960
I
I
1970
1980
-200
I
1990
I
1950
2t •00
1960
I
Yea/'
I
1980
I
1990
2000
Year'
lO0
i
I
1970
lOO
i
1
I
I
I
.
8o
'
4C. Temperate soil
•
80
60
ß•
60
40
¸
40
20
•
20
•
104
I
103
I
102
_<10•
Residence time, years
4D. Tropical
ß
104
Soil
t
103
102
_<10•
Residence time, years
Fig. 4. Typicalmodelresults(dashedlines)for the(a) temperate
and(b) tropicalsoils, 0-23 and0-22 cmlayers,
respectively.
Observed
14Cdatafor thetotaldensefractions
of thetwosoilsarealsoshown(points).The 1959
datafor thetropicalsoilsareshownasa rangeof acceptable
values,from-110 %0(observed
in thearchivedsoils)
to 0%0(theupperlimit for late 1950's14C). Dataandmodelresultsshownhererepresent
soilfractionswith
density
> 2.0g cm-3only.Thepartitioning
of carbon
among
the1-10year,100year,1,000yearand10,000year
poolsusedfor (a) thetemperate
soil(totalof 176gC m-2yr-1)was:(1) unevendashed
line,36 (20%),60 (34%),
80 (46%) and0 (0%); (2) unevendashedline (heavy),60 (34%), 56 (32%), 54 (31%) and5 (3%); (3) dashedline,
85 (48%),0 (0%),91 (52%),and0 (0%). Amounts
in thesamefourpoolsin (b), thetropicalsoil(425gCm-2crrr
• in 0-10 cm and333 gC m-2cm-•in 10-22cm)were:(1) unevendashed
line, 400 (94%),0 (0%), 0 (0%) and25
(6%); (2) unevendashedline (heavy), 155 (46%), 0 (0%), 150 (45%) and 28 (9%); (3) dashedand dottedline,
200 (60%),50 (15%),20 (6%), and63 (19%). (c) and(d) depicttherangeof permissible
valuesfor thesizeof each
ofthecomponents
(aspercentofthetotaldense
fraction
carbon).
These
ranges
werebased
onrepeated
model
runs,
during
whichanattempt
wasmadetomaximize
andminimize
thesizeofeachfraction
independent
oftheothers.
describethe cyclingof organicmatterin the0-23 cm
layer fully, the turnoverrate of the low-density
organicmatter must be determined. The best fit to
the 1990 datafor low-densitymaterial(p< 2.0 g cm3, exclusiveof charcoal)givesa residencetime for
this fraction of 20 years,assumingthis pool is at
steadystate and may be describedusing a single
decayconstant.
An alternative explanation for the observed
increases in bomb 14(2content in the two soils is that
the steadystateassumption
is not valid,andthesoils
havebeenaccumulating
carbonwith time. In order
to explainall of the 14Cincreasesobservedin the
soils (assuminga constantrate of carbonaccumulation in the soil since 1959), the total addition of
carbonto theAmazonsoil(0-22cmlayeronly)must
increasethe carboninventoryin this layer by 8090%. The requiredincreasein the temperatesoil
inventory (to 23 cm) is about 30%. The carbon
inventoryin the 1990 Ultisol profile is 22% lower
286
Trumbore:CarbonDynamicsin TropicalandTemperateSoils
(to 69 cm) thanthe 1959 archivedprofile(Table 2),
ruling out an accumulationhypothesisfor this soil.
The decreasein carbon content may reflect the
overall uncertainty of carbon inventory
determinations, site-to-site differences on meter
scales, or anthropogenicinfluences (cessationof
burning; cutting of vegetation). Because the
prebomband postbombtropical soils come from
differentsites,the accumulation
hypothesis,
though
producinga measurableeffect, cannotbe assessed
The datapresented
hereshowthatthefirst approach
underestimates the dynamic nature of SOM,
especiallyin tropicalregions. In addition,dramatic
differences in carbon dynamics are observed in
different soil orders, which may need to be
incorporated
into morecomplexglobalcarboncycle
models.
organicmatterin surfacesoils). More recentmodels
divide SOM into reactiveand refractorypools,but
Jenkinsonet al. [1991, 1992] point out that one of
the benefitsof a more quantitativeunderstanding
of
the soil carboncycle is in improvedestimatesof the
annualinputof carbonto soils. The annualinputof
carbonmay be estimatedfrom the amountof carbon
in pools with turnover times of _<10, 100, 1000,
and 10,000 years, and the assumptionthat the soil
carbonis at steadystate (see Figure 3). At steady
state,the annuallossof carbonfrom the soil equals
annual inputs and is the sum of the productsof
carboninventoryandturnovertime for thefive SOM
pools (including the low density fraction as a
separatepool). The calculationof inputsis shownin
little information
Table 3. Total fluxes are estimated as 0.22 to 0.45
here.
DISCUSSION
Global carbon cycle models have in the past
incorporatedsoil organicmatter (SOM) as a single
reservoirwith a residencetime of about 1000 years
(based on the bulk radiocarbon content of soil
is available with which to detmxnine
kgC m-2 yr-1 from the SierraNevadaUltisol, with
the amountof SOM which falls into eachcategory.
TABLE 3. Calculation
of InputRatesfor 0-22 cmand0-23 cmLayersof theTemperate
andTropicalSoilsFrom ModeledCarbonAbundanceandTurnoverRate.
Pool
DecayRate,
Amount,
Flux,
yr-1
kgC m-2
kgC m-2yr-1
TEMPERATE
p<2.0
0.05 - 0.10
2.6
1-10
100
0.1 - 1.0
0.01
0.9- 2.2
0- 1.3
1,000
10,000
0.001
0.0001
1.4 - 2.4
0 - 0.2
0.13 - 0.26
p<2.0
Total
2.6
0.22 - 0.45
TROPICAL
p<2.0
p<2.0
1.0
1.4
1-10
100
0.1 - 1.0
0.01
0.9- 8.8
0.9 - 2.2
1,000
10,000
0.001
0.0001
0.4 - 1.1
0.8 - 1.4
Total
10.9
1.4
2.6-
10.5
Turnoverratesfor 0<2.0g cm-3fractionsareestimated
from •4Cdata(seetext)andfrom
literaturevaluesfor litterdecomposition
for bothtemperate[Jennyet al., 1949]andtropical
[Klinge andRodriguez,1986] sites.The rangein valuesfor eachcomponentof the 0<2.0 g
cm-3fraction(fromFigures4c and4d) m'egivenin the"Amount"
column;individual
scenarios
like thosedepictedin Fugres4a and4b wereusedto calculatemaximumand
minimum
fluxes.
Trumbore:CarbonDynamicsin TropicalandTemperateSoils
half or more of the annual flux due to turnover of the
the Sierra Nevada site, and data from a site which is
within several km of the Reserva Ducke; see Table 4
low-density material in the 0-22 cm layer (total
fluxes include data to 60 cm).
Fluxes from the
for citations). Comparisonof estimatedannualC
fluxesinto andout of soil organicmatterare larger
tropicalsoil arelarger,0.5 - 4.1 kgC m-2yr-•. The
range in values is due to lack of sensitivity in
differentiatingcarbonpoolswith turnovertimesof 1
yearversus10 yearsin the•4C models.The 1977
than litterfall for both sites. The difference between
these two terms is a minimum estimate of the annual
contribution of root turnover to SOM;
overestimate the •4C content of the dense fraction
organicmatterin 1977. Figures4a and4b showthe
extremesensitivityof the modelingexerciseto •4C
part to the fact that litterfall estimateswere made in
an area more than 30 km away from the soil
samplingsite.
values in soils in the late 1960s and 1970s - archived
soilsobtainedfrom thisperiodwill havethe greatest
promiseof quantifyingorganicpoolsin soilswhich
over
on decadal
and
shorter
time
Soil CO2 flux measurementsmade during the
ABLE II experimentsare availablefrom the Reserva
scales.
Ducke site [Fan et al., 1989]. The measuredflux of
Estimatesof annualcarboninputsfor bothsoilshave
an additional 10 - 20% error reflecting the
uncertaintyin soilcarboninventory.
Table4 comparesinventoriesandfluxesin the soils
to other ecosystemC fluxes, suchas litterfall and
soil CO2 respiration,which may help constrainthe
annual
soil
carbon
fluxes
derived
from
CO2 was 1.3 to 1.5 kgC m-2 yr-•, with little
evidence of seasonal or diel variation.
An unknown
portion (usually assumedto be around 50% [D/3rr
andMiinnich, 1989]) of the soil respirationconsists
of root respiredCO2; the restis from decomposition
of soil organicmatter and root exudates. The CO2
flux data,whichincludeCO2 from decomposition
in
litter layers, suggestthat the fluxes due to soil
organicmatteroxidationmustbe at the lower end of
the estimatedfluxes given in Table 3, or closer to
•4C
modeling. The data on litterfall and litter turnover
are literature values determined in sites close to those
from which soil sampleswere taken (a site with
similar elevation and climate, and the same soil as
TABLE 4. Comparison
of Ecosystem
CarbonFluxesfor TemperateandTropicalSites.
Temperate
Tropical
INVENTORY,kgCm-2
Abovegroundbiomass
Litter
3.1 a
18.9 a
1.2 - 6.5 b
_
5.2
2.3
7.1
6.9
Soil A (0-22cm)
Soil B (22-60cm)
FLUX,kgCm'2 y-1
NPP C
0.3 - 0.5 a
1.0 a
0.06 b
0.36 - 0.40 c
0.01- 0.07b
0.22 - 0.45 d
0.01 - 0.02 d
_
0.5 - 4.1 d
_
-
1.3 - 1.5 e
Litterfall
directlitterdecomposition
InputsoilA
InputsoilB
CO2 flux
a Data arefrom Atjay et al., [1985].
b DataarefromJennyet al., [ 1949].
c DataarefromKlingeandRodrigues
[1968].
d Dataarefromthisstudy
e Data are from Fan et al., [1990].
the actual
contributionfrom roots dependson the degreeto
which organic matter is transferredfrom the litter
layer to the mineral soil. The tropical soil root
contributions
to soil organicmatterestimatedin this
way are equal to annual litterfall inputs. The
temperatesoil showsbelow groundinputsat least
four timeslitterfall;theselargevaluesmay be duein
data from Reserva Ducke suggestthat most of the
carbonin the 0 - 10 cm layer hasa residencetime of
1 year (see Figure 4b), which would suggestthe
higherend of the estimatedfluxes. However, these
samplesare purifiedsoil baseextractsandthusmay
turn
287
288
Trumbore:CarbonDynamicsin TropicalandTemperateSoils
0.5 kgCm-2yr-•. Onewayto resolvethedifference
is to measurethe •4C contentof CO2 respiredfi'om
Acknowledgments.
I would like to thankthe AMS
the soil.
where this work was conducted. My salary was
paid in Switzerland by a postdoc with W. S.
Broecker(Lamont-Doherty),and in Livermoreby a
post-doctoralappointmentat LLNL. Discussions
If the SOM is turning over on
predominantly
annualtimescales,
the•4Ccontent
of
respiredCO2 will trackatmospheric
14CO2.$OM
dominatedby decadalturnoverpoolsshouldhave
14CO2measurably
higherthanatmospheric
values.
Dependingon thecontributionof rootrespirationto
soil respiredCO2, thismay be bestobservedthough
incubation studies or direct field measurements.
CONCLUSIONS
Radiocarbonis an importantandunderutilizedtool
for quantifyingthe carbondynamicsof soil organic
matter and for testing models of soil carbon
dynamics.Previousproblemswhichforestailedthe
use of radiocarbon
in soils are now reduced due to
the availabilityof 14CAMS measurements.
The
comparison
of 14Cin prebombandpostbomb
soil
profiles for temperate and tropical ecosystems
demonstrates differences in the amount, character,
and turnoverrate of carbonin organicmatter. The
tropical forest soil containsmore carbon,which
turns over more rapidly, than the drier, temperate
forestsoil. Roughly50% of thecarbonin theupper
23 cm of the temperatesoil is low densitymaterial,
whilemostof theorganicmatterin theu'opica!
soilis
associatedwith mineral surfaces (dense fraction).
Hydrolysisof the densefractionwith 6N HC1left a
residuedepleted
in 14Cin thetemperate,
butnotthe
tropicalsoil.
Modelingof the14Cincrease
observed
in bothsoils
since 1963 demonstrates that the mean residence
time derived from bulk soil prebomb14C is a
misleading
indicatorof thedynamicnatureof thesoil
carbon.The tropicalsoilcontained
less14Cin 1959
thanthe temperatesoil,but showedgreaterincrease
in 14Csince1963. Modelingof the acidhydrolysis
residuefrom prebombandpostbombsoilsshowed
that this fraction cannot be used to define the amount
of "passive"
organic
matter,although
its14Ccontent
laboratories in Ztirich (PSI/ETH)
and at LLNL,
with W. S. Broecker, J. Harden, O. Chadwick and
R. Amundsonwerehelpfulto thecompletionof this
work.
R. F. Anderson at Lamont-Doherty
GeologicalObservatoryand J. Beer at EAWAG
kindlyprovided•37Csanalyses.BennyBrasherat
the U.S. Soil Survey InvestigationsLaboratory
kindly provided accessto archived soils. I am
indebtedto three anonymousreviewersfor helpful
comments.The writingof thispaperwasperformed
under the auspicesof the U.S. Department of
Energy by the Lawrence Livermore National
LaboratoryundercontractnumberW-7405-Eng-48.
REFERENCES
.•gren,G.I., andE. Bosatta,
Theoretical
analysis
of
thelong-termdynamicsof carbonandnitrogenin
soils,Ecology,68, 1181 - 1189, 1987.
Aravena, R., S. L. Schiff, S. E. Trumbore, and R.
Elgood, Evaluatingdissolvedinorganiccarbon
cyclingin a forestedlake watershed
usingcarbon
isotopes.Radiocarbon,34, 363-345, 1992.
Atjay, G. L., P Ketner and P. Duvigneaud,
Terrestrialprimaryproduction
andphytomass,
in,
The Global Carbon Cycle, SCOPE 13, editedby
B. Bolin, E. T. Degens, S. Kempe and P.
Ketner,pp. 129-182, John Wiley, New York,
1979.
Anderson,D. W., and E. A. Paul, Organomineral
complexesandtheirstudyby radiocarbon
dating,
Soil Sci. Soc. Am. J., 48, 298 - 301, 1984.
Amundson, R. G., and E. A. Davidson, Carbon
dioxide and nitrogenous gases in the soil
atmosphere,J.of Geochem.
Explor. 38, 13 - 41,
1990.
Balesdent,J., The turnoverof soil organicfractions
estimated by radiocarbon dating,Sci.Total
mayrepresent
a masonable
turnoverratefor passive
Environ. 62,405- 408, 1987.
pools.Annualratesof carboninputto soils,derived
Brady,N. C., The Nature and Propertiesof Soils,
frommodelingthe14Cincrease
in density-separated
pp. 415-473, Macmillan,New York, 1984.
soil organicmatter,showthatbelowgroundinputs
Buchanan, D. L., and B. J. Corcoran, Sealed tube
combustions for the determination of carbon-13
of carbonare at least as importantas litterfall in
addingcarbonto theA horizons
of bothsoils.
and total carbon, Anal. Chem. 31, 1635, 1959.
The abilityto differentiate
14Cincreases
overthe
Campbell,C. A., E. A. Paul, D. A. Rennie, and
past30 yearsdueto carbonaccumulation
fi'omthose
K.J. McCallum, Applicability of the carbondue to carbonexchangewith atmospheric
CO2 in
datingmethodof analysisto soil humusstudies,
steadystatesoilswill be important
if 14Cisotopes
can be used to determine
whether
soils
are
sequestering
carbonat an increasedrate due to
biosphere
fertilization.
At present,
14Cdatawill bea
necessarytest of any accumulationhypothesis.
Unfortunately,unlessthe soil carbondynamicsare
well understood
understeadystateconditions,
14C
dataalonearenot an independent
measureof carbon
sequestration
ratesin soils.
Soil Sci. 104, 217-224, 1967.
Davis, J. C., et al., LLNL/UC AMS facility and
research program: Proc. 5th International
Conference on Accelerator Mass Spectrometry,
Paris, Nuc. Instr. Methods, B52, 269-272, 1990.
DO•T,H., and K. O. Mtinnich, Downward transport
of soil organicmatter and its influenceon trace-
element transport(21øpb,137Cs)in the soil,
Radiocarbon,31,655-663,
1989.
Trumbore:
CarbonDynamics
in TropicalandTemperate
Soils
289
Emanuel, W. R., H. H. Schugart, and M.P.
Stevenson,Climate changeand the broad-scale
distributionof terrestrialecosystem
complexes,
Clim.Change,7, 29-43, 1985.
Fan, S-M, S.C. Wofsy, P.S. Bakwin, D. J.
Jacob, and D. R. Fitzjarrald, Atmospherebiosphereexchangeof CO2 and 03 in the central
Amazon forest, J. Geophys.Res., 95, 16,851 -
Jenny,H., Causesof high nitrogenand organic
mattercontentin temperateandtropicalsoils,Soil
16, 864, 1990.
Goh, K. M., T. A. Rafter, J. D. Stout, and T. W.
Leenheer, J. A., Origin and nature of humic
Walker, The accumulationof soil organicmatter
and its carbon isotope content in a
chronosequence
of soils developedon aeolian
sand in New Zealand,N. Z. J. Soil Sci. , 27, 89100, 1976.
Goh, K. M., J. D. Stout, and T. A. Rafter,
Radiocarbonenrichmentof soil organicmattter
fractions in New Zealand soils,Soil Sci., 123,
385-391, 1977.
Goh, K. M., J. D. Stout, and B. J. O'Brien, The
significanceof fractionationdatingin datingthe
age and turnoverof soil organicmatter,N.Z. J.
Science, 69, 63-69, 1950.
Klinge,H., andW. A. Rodriguez,Litterproduction
in an areaof Amazonianterrafirme forest,Part1,
Litter-fall, organic carbon and total nitrogen
contents of litter, Amazoniana,
1, 287 - 302,
1968.
substances in the waters of the Amazon
River
basin, Acta Amazon., 10, 513-526, 1980.
Levin, I., B. Kromer, H. Schoch-Fischer, M.
Bruns,M. Miinnich,D. Berdau,J. C. Vogel, and
K. O. Miinnich, 25 yearsof tropospheric14C
observations
in centralEurope,Radiocarbon,27,
1-19, 1985.
Manning, M. R., D.C. Lowe, W. H. Melhuish, R.
J. Sparks,G. Wallace,C. A.M. Brenninkmeijer,
and R. C. McGill,
The use of radiocarbon
measurements in atmospheric
Radiocarbon, 32, 37- 58, 1990.
Martel,
Y. A.,
studies,
and E. A. Paul, The use of
Soil Sci., 35, 69-72, 1984.
Graustein, W. C., and K. K. Turekian, 2•øpb and
•37Cs in air and soils measure the rate and vertical
radiocarbondatingof organicmatterin the study
of soil genesis,Soil Sci. Soc. A•n. Proc., 38,
profile of aerosolscavenging,J. Geophys.Res.,
Monaghan,M. C., •øBe in the atmosphere
and
91, 14,355- 14,366, 1986.
Harkness, D. D., A. F. Harrison, and P. J. Bacon,
The temporal distributionof 'bomb' •4C in a
forest soil, Radiocarbon, 28, 328-337, 1986.
Harriss.R. C., et al., The AmazonBoundaryLayer
Experiment:Wet season1987,J. Geophys.Res.,
95,16,721
- 16, 737, 1990.
Intergovernmental Panel on Climate Change,
Climate Change: The IPCC Scientific
Assessment,
UniversityPress,Cambridge,1990.
Irion, G., Clay mineralsof Amazoniansoils,in,The
Amazon, edited by H. Siolo, pp. 537-580, W.
Junk, Dordrecht, 1984.
Jawson, M.D.,
and L. F. Elliott, Carbon and
nitrogentransformations
duringwheat straw and
root decomposition,
Soil Biol. Biochem.,18, 1522, 1986.
Jenkinson,D. J., and J. H. Raynor,The turnoverof
soil organic matter in someof the Rothamsted
classical experiments,SoilSci.,123• 298- 305,
1977.
Jenkinson, D. J., D. E. Adams, and A. Wild,
Model estimates of CO2 emissions fi'om soil in
responseto global warming, Nature, 351,304306, 1991.
Jenkinson, D. S., D. D. Harkness, E. D. Vance, D.
E. Adams, and A. F. Harrison, Calculating net
primary productionand annualinput of organic
matter to soil from the amount and radiogenic
content of soil organic matter, Soil Biol.
Biochem., 24, 295- 308, 1992.
Jenny, H., S. P Gessel, and F. T. Bingham,
Comparativestudyof decomposition
of organic
matterin temperateandtropicalregions,SoilSci.,
67, 419 - 432, 1949.
501-506, 1974.
soils, Ph.D. thesis, Yale Univ., New Haven,
Conn., 1984.
O'Brien, B. J., Soil organic carbon fluxes and
turnover
rates
estimated
from
radiocarbon
enrichments, Soil Biol. Biochem., 16, 155-120,
1984.
O'Brien,
B. J.,Theuseof naturalanthropogenic
14C
to investigatethe dynamicsof soilorganiccarbon,
Radiocarbon, 28, 358-362, 1986.
O'Brien, B. J., and J. D. Stout, Movement and
turnoverof soil organic matter as indicatedby
carbon isotopic measurements, Soil Biol.
Biochem., 10, 309-317, 1978.
Parton, W. J., D. S. Schimel, C. V. Cole and D. S.
Ojima,Analysisof factorscontrollingsoil organic
matterlevels in Great Plainsgrasslands,
Soil Sci.
Soc. Am.J.,51,
1173-1179, 1987.
Paul, E. A., C. A. Campbell, D. A. Rennie and K.
J. McCallum, Investigationsof the dynamicsof
soil humusutilizingcarbondatingtechniques,
in,
Transactionsof the 8th InternationalSoil Science
Society (Bucharest, Romania), 3, 201-208,
InternaltionalSoil ScienceSociety,Vienna, 1964.
Post, W. M., W. P. Emanuel, P. J. Zinke, and A.
G. Stangenberger,Soil carbonpools and world
life zones,Nature, 298, 156-159, 1982.
Prentice,K. C., and I. Y. Fung, The sensitivityof
terrestrial carbon storage to climate change,
Nature, 346, 48 - 51, 1990.
Scharpenseel,H. W., C. Ronzani, and F. Pierig,
Comparative age determinationson different
humic-matter fractions, in, Proceedingsof a
Symposiumon the useof isotopesand radiation
in soil organicmatterstudies,July, 1968, pp. 67-
290
Trumbore:CarbonDynamicsin TropicalandTemperateSoils
74, International Atomic Energy Commission,
Scharpenseel,
H. W., M. A. Tamers,andF. Pietig,
Altersbestimmung von Boden durch die
Radiokohlenstoffdatierungsmethode,
Z. P.fianz.
Bodenk., I19, 34-52, 1968b.
Scharpenseel,H. W., H. Schiffmann, and B.
Hintze, HamburgUniversity radiocarbondates
III, Radiocarbon, 26, 196-205, 1984.
Scharpenseel,
H. W., P. Becker-Heidmann,
H. U.
Neue and K. Tsutsuki,Bomb-carbon,14Cdating
and •3C measurements
astracersof organicmatter
dynamicsaswell asof morphogenic
andturbation
processes,
Sci.Total Environ.,81/82, 99-110,
1989.
Schlesinger,
W. H., Carbonbalancein terrestrial
detritus. Annu. Rev. Ecol. Syst., 8, 51-81,
Stuiver,M., andH. Polach,Reportingof •4C data,
Radiocarbon, 19, 355-363, 1977.
Surer,M., et al., Precisionmeasurements
of •4C in
AMS - someresultsand prospects,Nl.tcl.Instr.
Methods, B5, 117-122, 1984.
Tans, P., A compilationof bomb 14Cfor usein
globalcarboncyclecalculations,
in, CarbonCycle
Modeling, SCOPE 16, edited by B. Bolin, pp.
131-157,John Wiley, New York,, 1981.
Tans, P., I. Y. Fung, and T. Takahashi,
Observationalconstraints
on the atmospheric
CO2
budget,Science,247, 1431-1438,1990.
Trumbore,S. E., Carboncyclingand gasexchange
in soils, Ph.D. thesis, Columbia Univ., New
York, New York, 1988.
Trumbore, S. E., J. S. Vogel, and J. R. Southon,
1977.
Schlesinger,
W. H., Evidencefrom chronosequence
studiesfor a low carbon-storage
potentialof soils,
Nature, 348, 232-234, 1990.
Soil Survey InvestigationsReport (SSIR), Soil
SurveyLaboratory
dataanddescriptions
for some
soils of Iowa, Rep. 3 (Iowa), Soil Conserv.
Serv., U.S. Dep. of Agri., Washington,D.C.,
1970.
SSIR, Soil SurveyLaboratorydataanddescriptions
for somesoilsof California,Rep. 24 (California),
Soil Conserv. Serv., U.S. Dep. of Agri.,
Washington,D.C., 1973.
Sollins,P., G. Spycher,andC. Topik,Processes
of
soil organic matter accretion at a mudflow
chronosequence,
Mt. Shasta,California,Ecology,
64, 1,273-1,282, 1983.
Sombroek, W. G., Amazon
A. Paul, and J. N. Ladd, Soil Biochem., 5, 173, 1981.
Vienna, 1968a.
AMS
14C measurements of fractionated
soil
organicmatter:anapproach
to deciphering
thesoi
carboncycle,Radiocarbon,
31,644-654, 1989.
Trumbore, S. E., G. Bonani, and W. W61fii, The
ratesof carboncyclingin severalsoilsfromAMS
14C measurementsof fractionated soil organic
matter, in, Soils and the Greenhouse Effect,
edited by A. F. Bouwman,pp. 405-414, John
Wiley, New York, 1990.
Van Breemen, N., and T. C. J. Feijtel, Soil
processes and properties involved in the
producitonof GreenhouseGases,with special
relevanceto soiltaxonomicsystems,
in, Soilsand
the Greenhouse Effect, edited by A. F.
Bouwman,pp. 405-414,JohnWiley, New York,
1990.
Soils, Centre for
Agricultural Publication and Documentation,
Wageningen,The Netherlands,1966.
Spycher,G., P. Sollins, and S. Rose, Carbonand
nitrogen in the light fraction of a forest soil:
Vertical distributionand seasonalpatterns,Soil
Sci., 135, 79-87, 1983.
Stevenson,F. J., Humus Chemistry, John Wiley,
New York, 1982.
Stevenson,F. J., and E. T. Elliott, Methodologies
for assessingthe quantity and quality of soil
organic matter, in, Dynamics of Soil Organic
Matter in Tropical Ecosystems,
editedby D.C.
Coleman, J. M. Oades, and G. Uehara, pp. 173242, University of Hawaii Press, Honolulu,
Vogel,J. S.,D. E. Nelson,andJ. R. Southon,
14C
background levels in an accelerator mass
spectrometry
system,Radiocarbon,
29, 323-333,
1987.
Zinke, P J Stangenberger,A G, Post, W M,
Emanual, W R and Olson, J S, Worldwide
organicsoil carbonand nitrogendata,Environ.
Sci. Div. Publ. 2212, Oak Ridge Nat. Lab., 141
pp., Oak Ridge,Tenn., 1984.
S. E. Trumbore, Center for Accelerator Mass
Spectrometry,LawrenceLivermore National
Laboratory,Livermore,CA 94550.
1990.
Stout, J. D., K. M. Goh, and T. A. Rafter,
Chemistry and turnover of natually occurring
resistantorganiccompoundsin soil, editedby E.
(ReceivedNovember 16, 1992;
revisedFebruary15, 1993;
acceptedFebruary25, 1993.)