332
0169-5347/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved.PII: S0169-5347(00)01906-6
TREE vol. 15, no. 8 August 2000
T
he inexorable rise of atmosphericCO
2
concentrations, as observed atMauna Loa since the 1950s, is perhapsthe most familiar scientific graph of the20th century
1,2
. It sits as indisputable evi-dence that human activity is modifyingthe earth’s atmosphere at a global scaleand is at the centre of the debate onglobal climate change. The cause of thisrise is well understood; CO
2
is being emit-ted through the large-scale burning of oil,coal and gas, which power modern indus-trial economies, with an additional con-tribution coming from the clearing oftropical forests and woodlands. How-ever, these changes are meshed withinan immense natural global carbon (C)cycle that is still poorly understood andthat will almost certainly provide newsurprises. Here, we focus on the role thattropical forests play in this global CO
2
exchange.Tropical forests occupy a broad bandgirdling the earth’s moist equatorialregions, occupying approximately17 560
10
3
km
2
in 1990 (Ref. 3). Usingthe definitions and estimates of the Foodand Agriculture Organization (FAO)
3
, theseregions include lowland evergreen rain-forests (7180
10
3
km
2
) at the equator,moist deciduous forests (5870
10
3
km
2
),dry deciduous forests (2380
10
3
km
2
), andhill and montane forests (2040
10
3
km
2
).In total, tropical forests and woodysavannas account for 50% of global forestarea. These biomes are characterized byhigh productivity – a recent model
4
esti-mated that the annual net carbon pro-duction of tropical forests is 18 Pg C (1 Pg
10
15
g) and that of savannas is 17 Pg C;in total, accounting for 60% of globalterrestrial photosynthesis.The principal components of theatmospheric CO
2
cycle can be summa-rized as follows:
E
ff
E
trop
atm
ocean
(
trop
other
)
5.5
0.5 1.6
1.0 3.3
0.2 2.0
0.8 1.8
1.6
The values shown in this equation arethe often quoted mean values for1980–1989 in Pg C year
1
; they representthe state of knowledge of the Intergovern-mental Panel on Climate Change (IPCC)in 1995 (Refs 2,5).
E
ff
is the rate of CO
2
emission through fossil fuel combustionand cement production,
E
trop
representstropical deforestation,
atm
is theobserved increase in atmospheric CO
2
,and
ocean
,
trop
and
other
are the netsequestration into the oceans, the tropi-cal terrestrial biosphere and other com-ponents of the terrestrial biosphere,respectively.
E
ff
and
atm
can be esti-mated with reasonable accuracy, but theother terms have greater uncertaintyattached; in particular, the terrestrialbiosphere sink has often been calculatedas a residue of the other terms. A greatdeal of research effort has gone intoquantifying these terms more effectively;here, we will focus on recent thinkingregarding the magnitude of
E
trop
and
trop
– the terms relevant to tropical forests.
A case study of the tropical forestcarbon cycle
The C dynamics of a forest are dominatedby the assimilation of CO
2
through grossphotosynthesis (
G
p
); the release of Cthrough autotrophic (plant) respiration,
R
a
; the transfer of C to the soil in the formof leaf, wood and root litter, and the exu-dation of organic compounds into therhizophere
6
; and the eventual release ofthis soil C back to the atmospherethrough decomposition and respirationby microbes and other heterotrophs(
R
h
). There are few tropical sites wherethe internal forest C cycle has been exam-ined in detail. Figure 1 shows a case studyfrom a dense lowland tropical forest,withno history of disturbance, near Manausin central Amazonia, Brazil.
G
p
and
R
t
were calculated directly from continuousmeasurements of CO
2
exchange abovethe forest canopy, using the microme-teorological technique of eddy covari-ance
7,8
. Recently, this technique has beenapplied to tropical forests
7,9,10
, and hasbegun to produce a flood of informationon the C balance of tropical forests andits relation to climatological variables.Of the
G
p
of 30.4 t C ha
1
year
1
,approximately 14.8 t C ha
1
year
1
arerespired through leaves, wood and roots;the remainder (15.6 t C ha
1
year
1
) istermed the net primary production,
N
p
.This C is fixed into plant structural bio-mass before eventually being depositedinto the soil in the form of litter, deadtrees or animal faeces, from where it iseventually released by microbial decay.The mean residence time of C in biomassand soil can be estimated by dividing therespective C stocks by
N
p
. The above-ground biomass stocks were measureddirectly (C is approximately 48% of drybiomass
11
), the belowground biomasswas assumed to be 33% of abovegroundbiomass
12
and soil organic C stockswere derived from other studies inAmazonia
13
.We estimated the mean C residencetimes to be 16 years in biomass and13 years in soils, totalling to 29 years
14
.Figure 1b shows a simplified version ofFig. 1a. There is a net transfer of CO
2
intobiomass at a rate of 15.6 t C ha
1
year
1
.This C remains in the biomass for a meanresidence time of 16 years, at which timeit is transferred to the soil where itremains for 13 years before beingreturned to the atmosphere.The value of
N
p
suggested hereexceeds the usual values of 6–14 t C ha
1
year
1
derived from field allometric stud-ies
15,16
. However, it is now widely recog-nized that almost all
N
p
estimates in theliterature are low because root turnoverand exudation have usually beenignored
17
. Aboveground growth and res-piration can account for only a fraction ofthe total
G
p
; there has to be a significanttransfer of C belowground to bring con-sistency to forestry and gas-exchangemeasurements. If fine root turnover andexudation are neglected in our calcu-lation, then
N
p
9.8 t C ha
1
year
1
–within the usual range. Is our estimate forbelowground C cycling realistic? The fewdirect studies in temperate trees
6
indi-cate that between 40% (for
Liriodendrontulipitera
) and 73% (for
Pinus sylvestris
) ofassimilated C is transferred belowgroundto roots and to mycorrhizae; in Fig. 1a itis 45% for a tropical forest, a fairly con-servative value compared with thesetemperate studies.Whatever their exact magnitude, itseems certain that belowground C flowsare a significant, and often underesti-mated, component of
N
p
. Thus, the value
PERSPECTIVES
Tropical forests and atmosphericcarbon dioxide
Yadvinder Malhi and John Grace
Tropical forests play a major role in determining the current atmosphericconcentration of CO
2
, as both sources of CO
2
following deforestation and sinks ofCO
2
probably resulting from CO
2
stimulation of forest photosynthesis. Recently,researchers have tried to quantify this role. The results suggest that both the carbonsources and sinks in tropical forests are significantly greater than previously thought.
Yadvinder Malhi and John Grace are at the Institute of Ecology and Resource Management, DarwinBuilding, University of Edinburgh, UK EH9 3JU (ymalhi@ed.ac.uk; jgrace@ed.ac.uk).
TREE vol. 15, no. 8 August 2000
333
PERSPECTIVES
of 18 Pg C year
1
, quoted earlier forglobal tropical forest productivity, mightbe an underestimate; 25 Pg C year
1
might be closer to the true figure.
Carbon emissions from tropicaldeforestation:
E
trop
The three principal zones of tropical for-est are in South America, central Africaand southeast Asia. All three areas areundergoing rapid deforestation, primar-ily because of clearing for croplands(55%) or cattle pasture (20%), or becauseof the expansion of logging and shiftingcultivation (12% each)
18
. Table 1 showsthe extent of tropical forests in 1990 (Ref.3), and the rates of clearance between1980–1990 and 1990–1995 (Ref. 19). TheAmericas account for over 50% of tropi-cal forest area, and the absolute clear-ance rates are highest here. However,relative clearance rates are highestin tropical Asia. Although they are stillhigh, deforestation rates appear to havediminished by 12% in the 1990s.Table 1 shows two recent estimatesof net C emissions from tropical land-usechange. Both estimates use data onannual land-use change, and stocks of Cin biomass and soil to estimate therelease of CO
2
at the time of ecosystemdisturbance. Houghton
18
attempted tocalculate annual C emissions from forestand soil, by tracking the decay andregrowth of C stocks in the years anddecades following disturbance; however,Fearnside
20
expresses C loss as net com-mitted emissions (the total amount of Clost as the landscape approaches a newequilibrium). If deforestation rates areapproximately constant, the two valuesare approximately equivalent.Houghton’s results are essentiallyan update of those used to provide theIPCC 1995 value for
E
trop
of 1.6
1.0 Pg Cyear
1
, but they now include a re-vised and detailed analysis of tropicalAsia
21
, which has increased the estimateof mean 1980s emission in that regionfrom 0.7
0.3 to 1.1
0.5 Pg C year
1
.Fearnside’s study is centred on a detailedanalysis for Brazilian forest andsavanna
12
, which is based on a regionalevaluation of C stocks, field studies offorest burning and satellite-deriveddeforestation rates.These new studies significantlyincrease the estimated 1980s value of
E
trop
to about 2.0 Pg C year
1
, but in dif-ferent regions and for varied reasons. Inthe final column in Table 1, we suggest acurrent ‘best estimate’ for C emissions inthe 1980s, by combining the new ana-lyses for the Americas
20
and Asia
21
, andaveraging the results for Africa. This nowpushes the estimated emissions fromtropical deforestation to 2.4
0.6 Pg Cyear
1
. However, even this might be anunderestimate, because several recentstudies have suggested that humanactivity induces significant biomassreduction in apparently intact foreststhat is not accounted for in current analy-ses. Fearnside estimated that a further0.4 Pg C year
1
are emitted through for-est degradation. A field survey
22
of woodmills estimated that forest impoverish-ment through logging contributes an addi-tional 4–7% to the net C release throughAmazonian deforestation, and leaves the
Fig. 1.
(a)Carbon (C) fluxes and stocks in a dense tropical rainforest near Manaus, Amazonia, Brazil(after Malhi
et al.
14
).
D
AG
, aboveground detritus;
D
BG
, belowground detritus;
G
p
, gross primary pro-ductivity;
R
a
, autotrophic respiration;
R
h
, heterotrophic respiration,
R
l
, leaf respiration;
R
r
, root respi-ration;
R
t
, total respiration,
R
w
, aboveground wood respiration; and
T
, belowground C translocation.Fluxes of C as CO
2
are shown with grey arrows and fluxes of C as solid organic matter are shownwith black arrows. Stocks of C are shown in bold:
AG
, C in aboveground biomass;
AG
, annual incre-ment in
AG
;
BG
, C in belowground biomass;
BG
, annual increment in BG;
SOM
, C in soil organicmatter; and
SOM
, annual increment in
SOM
. Units are t C ha
1
year
1
for fluxes and t C ha
1
forstocks.
Reproduced, with permission, from Ref. 14
. (b) A simplified C flow diagram for the same for-est.
N
p
, net primary productivity;
R
h
, heterotrophic respiration;
SOM
, C in soil organic matter; and
mean C residence times (years) in biomass and in soil. All terms, except
G
p
and
R
t
, were extrapo-lated from litterfall, forestry measurements and gas-exchange measurements, or were estimated.See Malhi
et al.
14
for more details of the measurements and assumptions involved.
Trends in Ecology & Evolution
(b)
(a)
G
p
30.4
R
t
24.5
AG
+ 1.7
D
AG
7.0
T
13.7
R
a
14.8
R
h
9.7
N
p
15.6
R
l
4.1
R
w
3.9
R
r
6.8
D
BG
6.3AG = 180
BG
SOM
+ 3.6BG = 64SOM = 162
+ 0.6
Biomass = 244= 16 years
R
h
SOM = 162= 13 years
Table 1. Deforestation rates and estimated resulting CO
2
emissions in tropical areas
a
RegionForest area
b
Moist forest area
c
Deforestation rate
a
Deforestation rate
a
Carbon emissionCarbon emissionCarbon emission(10
3
km
2
)(10
3
km
2
)1980–19901990–1995Houghton
a
Fearnside
a
combined
d
(10
3
km
2
year
1
)(10
3
km
2
year
1
)(Pg C yr
1
)(Pg C yr
1
)(Pg C yr
1
)Americas9179.28705.164.856.90.550.940.94Africa5275.93730.241.937.00.290.420.36Asia3106.02663.739.735.11.080.661.08
Total17561.015098.9146.4129.01.902.002.40
a
Deforestation rates 1980–1990 and 1990–1995, taken from Ref. 18. Houghton
’
s estimates of carbon (C) emissions taken from Ref. 19 and Fearnside
’
sestimates of C emissions taken from Ref. 20.
b
Incorporates rainforest, moist deciduous forest, dry and very dry forest, hill and mountain forest, and alpine area forest, as defined by the Food and AgricultureOrganization (FAO)
3
.
c
Incorporates only rainforest, moist deciduous forest, and hill and mountain forest, as defined by the FAO (Ref. 3).
d
Estimate for C emissions in the 1980s by combining the new analyses for the Americas
20
and Asia
21
, and averaging the results for Africa.
334
TREE vol. 15, no. 8 August 2000
PERSPECTIVES
forest vulnerable to ground fires
23
thathave the potential to more than doubletotal C emissions in dry years. In addi-tion, much of the surviving forest, adja-cent to areas of deforestation, is frag-mented
24
and, thus, prone to biomassloss even in the absence of humanharvesting
25
.These recent studies suggest that it isnot implausible that the true rate of CO
2
emission through tropical deforestationin the 1980s might have approached3.0 Pg C year
1
, almost double the valuereported by IPCC in 1995 (Ref. 5). If so,one question begs an answer: where isthis extra C going?
CO
2
in the tropical atmosphere
An independent view on terrestrial Cemissions can be obtained by examiningthe temporal and spatial variation of CO
2
concentrations in the atmosphere.Recent studies have attempted to pro-vide continental-scale maps of the distri-bution of C sinks and sources
26–28
. Thereare small gradients in mean atmosphericCO
2
concentration between hemispheresand between continents, caused primar-ily by the uneven spatial distribution ofsurface C sources and sinks, and by rela-tively slow mixing by the troposphere. Inprinciple, if the global distribution of CO
2
concentrations and fossil fuel emissions,and the global transport and dispersionrates estimated from meteorologicalmodels, is known, the models can beinverted to provide a spatial surface mapof CO
2
sources and sinks. Monitoring theratios
13
C:
12
C and O
2
:N
2
can provide fur-ther information on the relative partitionof the fluxes between oceans and terres-trial biosphere
29,30
. However, there areseveral methodological difficulties withthis approach
14
, including insufficientlong-term CO
2
concentration records, dif-ficulties in assessing tropospheric CO
2
concentrations from near-surface meas-urements
31
and the problem being poorlyconstrained mathematically. The tropicsare a particularly difficult region in whichto carry out such an analysis becausethere are few CO
2
sampling sites
32
, andbecause the intensity of tropical convec-tion results in vigorous mixing and dilu-tion of any spatial patterns of CO
2
con-centration
31
. Such problems shoulddiminish as the distribution and accu-racy of CO
2
sampling stations improves.Table 2 presents results from tworecent analyses of atmospheric CO
2
distribution. Rayner
et al.
27
used globalCO
2
measurements over the period1980–1995, and a long-term recordof
13
C:
12
C and O
2
:N
2
. Bousquet
et al.
28
analysed CO
2
concentrations over theperiod 1985–1995.
.
The two results dis-agree significantly both in the spatial dis-tribution of CO
2
sources and sinks, and inthe net tropical balance. Rayner
et al.
27
suggest a modest net sink in the terres-trial tropics, whereas Bousquet
et al.
28
suggest a significant net source concen-trated in southeast Asia. There are stillseveral inconsistencies with thisapproach.Commentaries on these results usu-ally focus on the C sink in temperateregions. Perhaps this reflects the geo-graphical bias of the scientific researchcommunity. Also, it is not always empha-sized that these atmospheric techniquesshould find a large CO
2
source in tropicallatitudes, of the magnitude indicated bythe deforestation studies – neither ofthese studies does so. This featureencapsulates the debate surrounding theC balance of tropical forests: studies ofdeforestation indicate that tropicalregions are releasing large quantities ofCO
2
,but only some, or none, of this CO
2
ismeasured in the tropical atmosphere.There are several possible reasons forthis discrepancy:•The deforestation studies stronglyoverestimate the net C release accom-panying land-use change.•The atmospheric transport modelsunderestimate the mixing of CO
2
between low and mid-latitudes, and,thus, underestimate how quickly thetropical CO
2
signature is dissipated.•There is a large CO
2
sink in the terrestrialtropical biosphere.Working on the third assumption andusing the ‘best estimate’ of tropical defor-estation in Table 1, the third and fourthcolumns in Table 2 show the magnitude ofthe C sink that would be required in eachregion to bring consistency betweendeforestation and atmospheric studies.The implied sink is significant and isprobably distributed across all threecontinents. Is such a C sink plausible?
A tropical carbon sink:
trop
The net C budget of an area of forest isthe balance between net production andheterotrophic respiration. As shownin Fig. 1b, these two terms must bein approximate balance because anychange in
N
p
eventually produces a cor-responding change in
R
h
, with a lag timeequal to the sum of the soil and the bio-mass residence times. For example, ifthere is a short-term increase in
R
h
,because of an increase in soil tempera-ture, the soil C stocks will eventuallydecrease to bring
R
h
back to a level with
N
p
.However, variations in
N
p
and
R
h
on atimescale shorter than the residencetime can result in a net flux of C to orfrom the forest system. Thus, interannualvariation in cloudiness, precipitationand temperature might result in signifi-cant interannual variations in forest Cbalance
33
.There is also probably a longer termshift superimposed on this interannualvariation. Mean atmospheric CO
2
con-centration has increased from a prein-dustrial value of 280 ppm to 366 ppm in1998 (Ref. 1), and is currently increasingat about 1.5 ppm year
1
. A recent reviewof experimental studies growing treesin open-top chambers
34
indicates that a300 ppm increase in atmospheric CO
2
concentration stimulates photosynthesisby 60%, the growth of young trees by 73%and wood growth per unit leaf area by27%. It seems probable that there will bea similar response in natural forestecosystems. Because of their intrinsichigh productivity, tropical forests are aprime candidate for such a C fertilizationresponse
35
; the crucial question has beento what extent such a response might belimited by low nutrient availability, inparticular by low nitrogen or low phos-phorus. Several studies have attemptedto model the effect of rising CO
2
on tropi-cal forest productivity, with the magni-tude of the resulting increase beinglargely dependent on how the nutrientcycle is modelled
36,37
. Recently, it hasbeen argued that forests might simply
Table 2. Net carbon (C) balance and implied biotic C sink intropical forest regions
a
Net balance
a
Net balance
a
Net sink
b
Net sink
b
Rayner
et al.
Bousquet
et al.
Rayner
et al.
Bousquet
et al.
Region(Pg C year
1
)(Pg C year
1
) (Pg C year
1
) (Pg C year
1
)Americas
0.20.1
1.1
0.8Africa
0.00.2
0.4
0.2Asia
0.10.8
1.2
0.3
Total
0.31.1
2.7
1.3
a
The net C balance as inferred from analysis of atmospheric concentrations of CO
2
. The Rayner
et al.
estimate is taken from Ref. 27; and the Bousquet
et al.
estimate of net C balance is taken from Ref. 28.
b
The implied biotic C sink, calculated by subtracting our best estimate of CO
2
emissions from defor-estation (last column of Table 1) from the estimates of net C balance (Columns 2 and 3, above).
Search History:
Result 00 of 00
00 results for result for
p.
More From This User
Notes
Load more
