99, NO. D8, PAGES 16,421-16,427, AUGUST 20, 1994. Carbon monoxide fluxes from natural, managed, or cultivated savannah grasslands. Eugenio Sanhueza ...
JOURNAL OF GEOPHYSICAL
RESEARCH, VOL. 99, NO. D8, PAGES 16,421-16,427, AUGUST 20, 1994
Carbon monoxide fluxes from natural, managed, or cultivated savannah grasslands EugenioSanhueza and Loreto Donoso InstitutoVenezolanodeInvestigaciones Cientificas,Atmospheric ChemistryLaboratory, Caracas, Venezuela
Dieter Scharffe and Paul J. Crutzen Max-Planck-Institut far Chemic, Luftchemie, Mainz, Germany
Abstract.As partof a comprehensive studyon tropicallandusechangeandits effect on atmospheric tracegasfluxes, we reportthe CO fluxesrecordedat a natural grassland siteandthe changes produced whenthisecosystem wasmanagedor cultivated.The field site is locatedin the centralpart of the savannahclimaticregion of Venezuela.Fluxeswere measuredin the dark usingthe enclosedchamber
technique. CO wasanalyzedwith a reduction-gas detectorin combination with a molecularsieve5A colummfor CO separation. At all sites,CO fluxesexhibiteda strongdiurnalvariation,with net emissionduringdaytimeandconsumption or no fluxesduringnighttime.In unplowedsoilsno differences wereobserved betweendry andrainyseason. A largedisparitywasobserved betweenunplowedandplowed grassland soils.Ploxved soil showsa muchsmalleremissionduringdaytimeanda largerconsumption at night.The 24-hourintegratedfluxesindicatethatthe
nonperturbed grassland switches frombeinga netsource of CO(3.4x 10•ømolecules cm-2S-1)to beinga netsink(-1.6x 10•ømolecules cm-2S-1)afterplowing. It is likely
thatburialof surfacelitterreducesthe production of CO in the top soil andthatthe diffusionof CO to deeperlayers(whereCO is consumed by microbiological processes) is promoted in decompacted soils.As therainyseason progressed the plowedsoilgraduallycompacted andCO fluxeschanged back,andafter3 monthsthe fluxesfromplowedsoilsandthe originalunplowedsoilswereequal.Eventhoughthe variouscultivatedfields(corn,sorghum,andpasture)receiveddifferinginorganic fertilizationtreatments, no significantdifferencein the CO fluxesresulted. Measurements duringthe d• seasonsuggest that"degrading dry.(dead)vegetation" producesCO underdark conditions. 1. Introduction
Carbonmonoxide(CO) is a reactivegasthat affectsthe oxidizing efficiency of the atmosphere[Cn•tzen and Zimmermann,1991], and thereforethe concentrationof manygasesthat are vitally importantto tropospheric as well as stratospheric chemistry.At present,the chemical compositionof the atmosphereis changingrapidly [IntergovernmentalPanel on Climate Change (IPCC), 1992], resulting mainly from increasing industrial emissionsand changesin land use. Most sourcesof anthropogenic emission are locatedin the midlatitudes of thenorthernhemisphere, whereasmostchanges in landuse are occurringin the tropics due to deforestationand conversionof native savannahecosystems to agricultural
continentalsiteswhere anthropogenicemissionsare low (i.e. tropicalAmerica).Tropicalsavannahs covera large area of the continental world. However, little is known
aboutthegasexchange betweenthesoil-vegetation system andthe atmosphere. As partof a comprehensive research
projecton tropicalland use changeand its effect on atmospheric trace gas fluxes,we reportthe CO fluxes producedfrom a naturalgrasslandsavannahand the changesobserved whenthis ecosystem wasmanagedor cultivatedwith corn(Zea maysM.), sorghum(A ndropogon sorghum),and pasture(Brachiaradecumbens). 2.
Field
Measurements
The studywasperformedat theEstaci6nBio16gicadelos LLanos (BiologicalStationof the Plains),Gufiricostate, CO hasa relativeshortatmospheric lifetime(2-3 months) Venezuela (8ø532q;67ø19'W), which is located in the andthesoilvegetationsystemcouldbe an importantfactor central part of the savannah climatic region. Two in determiningthe local or regional CO budget at well-definedclimatic periodsoccur in the area: a dry seasonfrom Decemberto April and a rainy seasonfrom May to November. The annual rainfall is 1300 mm Copyright1994 by the AmericanGeophysicalUnion. (22-yearaverage);theannualmeantemperature is 27.6 øC, land.
Papernumber93JD02918.
with a maximum mean of 33.2 øC, and a minimum mean
0148-0227/94/93JD-02918505.00
of 22.9 øC. The soils and vegetation of the natural 16,421
16,422
SANHUEZA
ET AL.: CO FLUXES FROM TROPICAL GRASSLANDS
scrub-grass savannahhave been describedby Lourido and Bastardo[1986]. The soils are acidic, with a low rate of mineralization and poor in nutrients. They support gramineagrasses(mainly Trachypogonsp. andA xonopus canescens) interrupted by trees and scrub (Curatella americana, Boudichia crassifolia).
virgilioides
and
Byrsonima
At the managedsite, the trees and scrubwere removed and the grass cut regularly. Cultivation was carried out following local practice; the grasslandsoil was plowed (usinga rototiller) severaltimes betweenMarch and May. The corn was planted in rows with 1-m distancebetween them, and-0.6 m between plants in the same row; the fertilizer
was added over a small area around the seeds or
plants. In the pasture and sorghum fields, seeds and fertilizer were "distributed"homogeneously over the entire field. Dates of cultivation
events and amounts of fertilizer
used are given in Table 1. At the cornfield, flux measurementswere made from two types of locations (plots): one directly over the site where a corn plant was growing(wherefertilizer was added)andthe otherbetween the com rows ("bare soils",where no fertilizer was directly added);at the sorghumfield, measurements werealsomade from locations "with" and "without" plants. When necessary,corn and sorghumplants were clipped to the
heightof the chamber;plotswere changedperiodicallyand "fresh"plantswere cut every new measurement period.At the varioussitesthephysicalandchemicalpropertiesof the soilswere obtainedfrom soil samplesof 10-cmdepth,and the relevantresultsof the analysisare given in Table 2. A complete characterizationof the soil has been given by
Table 1. Cultivation Events During 1991.
Cdrdenaset al. [1993]; the variations observedbetween
unplowedand plowed soilsare presentedin a companion paper [Sanhueza et al., this issue]. Gravimetric soil moistureswere measuredevery 3 hoursin samplesof 2-cm depth. Soil temperature at 1-cm depth was recorded continuouslywith a thermocouple. Fluxes were measured using the enclosed chamber technique; the stainless steel-glasschamber and other details were similar to those described by Conrad and Seller[1985]. In this study,the chamberwas permanently covered
Field
Event
March 20 and 27 all
plowing
April 10 and 24
all
plowing
May 2
corn
seeding/NPKfertilization a
May29
pasture
seeding
June2
corn
germination
June 10
pasture
germination
June 18
corn
urea fertilization b
June 18
pasture
NPK fertilizationc
June21
corn/pastureherbicideaplication
June22
corn/pasture insecticideaplication
June24
sorghum
plowing
June25
sorghum
seeding/NPKfertilization a
July 10
sorghum
germination
July 19
sorghum
urea fertilization
NPK, nitrogen,phosphorusand potasium.
a50mg NPK/cm2;NPK: NH4+-N12%, pO43'-P24%, KCI-K
12%.
43 mg Urea/cm 2. 3.4 mgNPK/cm2.
aluminum
foil.
Carbon
monoxide
chamber at a flow rateof 250 mL min-•. The gasstream passedthrougha cold trap at -78 øC to removewater. A particularplot was measuredfor 60 min, duringwhich the following samples were injected to the gas chrommatograph: onesampleof ambientair, sevensamples from the chamber(1 every 6 min), and two samplesof calibrationgas.To initiate measurement,the chamberwas placed manually over the frame; after the measurements have been taken (-60 min) the chamber was removed to another frame and the process repeated. High-pressure mixturesof CO in syntheticair (258 ppb) were usedfor calibration
in the field.
These
calibration
mixtures
were
prepared by static dilution from a commercialmixture suppliedby Deuste-Steininger(Muhlhausen,Germany), stored for several months to reach equilibrium, and calibratedwith a dynamicdilution of CO in nitrogenusing a tunable-diodelaser as measuring system.Blank tests showingthat neither the chambernor the tubingmaterial interfere
Date
with
concentrationswere measured using a reduction-gas detector(traceanalytical)in combinationwith a molecular sieve 5A column for CO separation.Gas sampleswere automaticallysuppliedto the injection valve by a metal bellow pump (model MB-21) and recirculatedback to the
with
the
CO-flux
measurements
were
made
previously[Scharffe et al., 1990]. Most of the experimentsshowthat after the chamberwas closedvariations(increaseor decrease)in the concentration were observed.After a relatively shortperiod,changesin concentrations decline (eventually an equilibrium concentration was reached), and fluxes (emission or consumption)were calculatedusing only the initial points (at least three) of a particular run [Scharffeet al., 1990]. 3.
Results and Discussion
Soil is recognizedas a sink for atmosphericCO, it is estimated to remove 250-390 TgCO/yr [IPCC, 1992]. However, net productionof CO has been observedfrom arid subtropical[Conrad and Seller, 1982] and tropical savannah[Scharffe et al., 1990] soils. The net flux of CO from soils is determined by the competition between biogenic consumption, which is proportional to the atmosphericconcentrationof CO, and its productionfrom chemical reactionsin soil, which is independentof the atmosphericCO concentrationand depend on the soil organicmatter content;both processesare affectedby the moisture and the temperatureof the soil [Conrad and Seller, 1980, 1985; Scharffe et al., 1990]. On the other hand,it seemsthat higher plantsdo act as a net sourceof atmosphericCO, estimated at •100 TgCO/yr. Various reaction mechanisms have been proposed for CO production,the photooxidationof plant cellular material
SANHUEZA
Table 2.
ET AL.'
CO FLUXES
FROM
TROPICAL
GRASSLANDS
16,423
Soil Characteristics
Site
Period
Type
pH,
Organic Matter,
KCI Natural Grassland
Managed b Grassland
Cultivated•,• Grassland
%
Oct-Nov, 1990
sandy-loam
4.3 _+0.1(5)•
1.4 _+0.1(3)•
wet season dw season
sandy-loam sandy-loam
3.8 _+0.2(17) 4.0 _+0.1(4)
1.7 _+0.5(17) 1.4 _+0.3(4)
May 29 to June 10 June 11 to July 21
sandy-loam sandy-loam sandy-loam
3.7 _ 0.1(16) 3.8 _+0.2(25) 3.9 _+0.1(4)
1.9 _+0.6(17) 1.7 _+0.3(25) 1.4 _+0.2(4)
Jan 30 to Feb 7
aAverage+ sd (number of measurements). b1991-1992.
½Valuesfrom corn, sorghum,and pasturefields.
beingthe mostlikely; thereforelight is an importantfactor influencing CO emission, and laborator3•and field experimentsindicatethat emissionincreasedlinearlywith the irradiation intensity and that no CO productionis observed in the dark [Seller et al., 1978; Bauer et al., 19791. The main purpose of this study was to evaluate the changes in the CO soil fluxes that occur when the grasslandis managedor cultivated.Thereforeto evaluate the effects of roots in active plants, measurements were
madewith the intact("unperturbed")soil-grasssystem.To avoidthe photoproduction of CO from the plantsandalso the possible secondary production of CO due to photooxidationof reactivehydrocarbons, the chamberwas maintainedcompletelycoveredduring the measurement period. 3.1.
Natural
3.2. Managed Grassland At this site, trees from the original grasslandwere removed(morethan 10 yearsago) and grasswas regularly cut. The soil characteristics of both natural and managed grasslandare very similar (see Table 2). In comparisonwith the natural savannah, much higher emissionsof CO were observedat this site duringdaytime
(,upto 9x101ømoleculescm-2 s-1) and practicallyno consumption of atmospheric CO wasobservedat nighttime (see Figure 2). Since here the measurementswere made with an intact soil-vegetationsurfaceand the vegetation cover is more even than at the more patchy natural site, 120
UNDISTURBED
IOO-
Grassland
Average soil fluxes recordedas a function of the local time are given in Figure 1; fluxes obtained from plots where grasswas clippedto just above the soil surfaceare also includedin this figure. Emissionof CO is observed duringdaytime,whereasconsumption of atmosphericCO occurs during nighttime.Similar resultswere observedat another natural savannahsite [Scharffe et al., 1990], and the most likely explanationfor this diurnal variation is a strongtemperaturedependence of the chemicalproduction of CO. To obtain the net daily flux, it is necessaryto integratethemeasuredfluxesover24 hours.Usingthe data given in Figure 1, we obtain net averagefluxes of 8.1 x
•T __• -20o
E
-40
0 CLEARED
108molecules cm'2 s4 and -3.6 x 109molecules cm'2 s'1 for
unperturbedand perturbedgrasslandsoil, respectively.At present, we do not know the cause of the difference observedbetweenboth typesof plots; as far as we know, vegetationdoesnot produceCO in dark conditions[Seiler o!oo o!oo ,:oo ,!oo o;oo and Conrad, 1987]; we can speculatethat when the grass L ocol time was cut and the surface of the plot "cleared", some organicmatter (i.e., plant debris,litter, which by chemical Figure !. Diurnal variation of the CO fluxes from natural decompositioncould produceCO) was alsoremoved.We savannah. Each point is the average of several discuss later the possible production of CO from measurements madeat threedifferentplots.Error barsare standard deviations. degradationof dry vegetation.
-20
•
I
[
I
I
I
•
I
16,424
SANH•ZA
ET AL ßCO FLUXES
FROM TROPICAL
GRASSLANDS
180
14.0
n Roiny
-
120
-
1 oo
-
seeson
X Dry season
--
--
--
80--
60--
4.0-
20-
m
--2.0
-
--4.0
-
X
I
o
•
I
I
I
4
I
I
8
I
I
12
Local
I
I
16
I
20
24
time
Figure 2. Diurnalvariationof the CO fluxesfrom managedgrassland. Eachpointis the averageof severalmeasurements made at two differentplots. Error barsare standarddeviations.
it is likely that the production of CO from the decomposition of a largeramountof dryvegetationpresent at the surfaceof the managedgrasslandsoil causesthe higher emissions. Surprisingly, as we seein Figure2, the fluxesobtained duringthe dry seasonare very similarto thoserecorded duringthe wet season.Sincesoilmoistureandvegetation activitygreatlycontrastbetweenthe two seasons, we think that this result is fortuitousand doesnot signify that the fluxes do not depend on these factors which typically
facilitated, which, as we discuss in the next section,
increasesthe consumption of CO within deeperlayersof the soil. On the other hand, higher soil temperaturesare observedduring the dry season(see Table 3). Also, the presenceof a "large"amountof dry grassin the chamber during the dry seasonmeasurementscould produce a significantamountof CO by "degradation" (seedry season section).It is likely that a different combinationof these factors results in similar 3.3.
Cultivated
fluxes in both seasons.
Grassland
Soils
control soil fluxes. Under dry soil conditions,production and consumption rateswithin the soil shouldbe depressed Agriculturalpracticesimposedon "natural"ecosystems in soils.In the shortterm, the [Conrad and Seiler, 1985] but transport processes producemajor disturbances
Table 3. Twenty-Four-HourIntegratedFluxes,Obtained From Diurnal VariationsGiven in Figures1,2, and 4 Site
Period
Natural Grassland
Oct 23 to Nov 8, 1990b
Managed Grassland
Cultivated • Grassland
CO Fluxes. Soil Moisture, moleculescm-es-• %
Oct 23 to Nov 8, 1990• wet season, 1991 dry season,1992
May 29 to June 10, 1991 June 11 to July 21, 1991 Sept 28 to Oct 9, 1991 Jan 30 to Feb 7, 1992
Soil Temperature. oC
-3.6x109
2-10/4.4 (13)*
3.4x10lø 2.8x10•ø
1-21/6.3 (218)
26-30; 27 27-42; 33
-l.6x10•ø 5.3x109 2.1x10•ø 1.7x10•ø
1-18/3.2 (1 1-21/9.0 (179 2-24/10 (138)
