Carbon stock assessment for a forest-to-coffee conversion landscape in Sumber-Jaya (Lampung, Indonesia): from allometric equations to land use change analysis

Abstract

The change in stored carbon (C) stocks was assessed for a 700 km 2 area where forest cover decreased from 60% to 10% in the last 30 years. At the same time, the area under coffee increased from 7% to 70% with a gradual evolution from open "sun coffee" systems to multi-strata "shade coffee" systems that provide a partial compensation for C loss. The use of a generic tropi- cal forest rather than tree-specific allometric equation can lead to substantial (up to 100%) overes- timates of aboveground biomass depending on wood density and tree shape. The shoot:root ratio (biomass) of coffee shifted with age, from the 4:1 value often assumed for tropical trees to 2:1. Annual aboveground C stock accumulation rates during the establishment stage after slash-and- burn land clearing were 1, close to 2 or 3.5 Mg C ha−1a−1 for sun coffee, shade coffee and fallow regrowth, respectively. Forest remnants, shade coffee and sun coffee had soil C stocks in the up- per 30 cm of the soil that were 79%, 60% or 45%, respectively, of the values expected for primary forest in Sumatra. Total C stock (time averaged, above - 0.3 m in the soil) for forest, shade and sun coffee was 262, 82 and 52 Mg C ha−1, respectively. In the 19701984 period, while forest cover was reduced from 59.5% to 19.7%, the landscape lost on average 6.8 Mg C ha −1 a −1 .I n the 19842000 period forest cover was further reduced to 12.6%, but the landscape lost only 0.39 Mg Ch a −1 a −1 , as forest loss was partially compensated by an increase in shade coffee systems. Conversion of all current sun coffee to shade coffee systems while protecting the remaining forest, could increase average landscape level C stocks by 10 Mg ha−1 over a time frame of say 20 years, or 0.5 Mg C ha−1 a−1.

Full text

1
Carbon stock assessment for a forest-to-coffee conversion landscape in
Sumber-Jaya (Lampung, Indonesia): from allometric equations to land use
change analysis
Meine van Noordwijk
1
, Subekti Rahayu
1
, Kurniatun Hairiah
2
, Wulan, Y.C.
1,2
, Farida
1
, Bruno
Verbist
1
International Centre for Research in Agroforestry (ICRAF) SE Asia, P.O.Box 161, Bogor 16001,
Indonesia; m.van-noordwijk@cgiar,org
Brawijaya University, Malang, Indonesia
Bogor Agricultural University, Bogor, Indonesia
Abstract
The change in stored carbon (C) stocks was assessed for a 700 km
2
area where in the last
30 year forest cover decreased from 60 to 10% while the area under coffee increased
from 7 to 70%, but where a gradual evolution from open ‘sun coffee’ systems to
multistrata ‘shade coffee’ systems provides a partial compensation for C loss. Use of a
generic tropical forest rather than tree-specific allometric equation can lead to substantial
(up to 100%) overestimates of aboveground biomass depending on wood density and tree
shape. Shoot:root ratio (biomass) of coffee shifted with age, from the 4:1 value often
assumed for tropical trees to 2:1. Annual aboveground C stock accumulation rates during
the establishment stage after slash-and-burn land clearing were 1, close to 2 or 3.5 Mg C
ha
-1
year
-1
for sun coffee, shade coffee and fallow regrowth, respectively. Forest
remnants, shade coffee and sun coffee had soil C stocks in the upper 30 cm of the soil
that were 79, 60 or 45%, respectively, of the values expected for primary forest in
Sumatra. Total C stock (time averaged, above –0.3 m in the soil) for forest, shade and sun
coffee was 262, 82 and 52 Mg C ha
-1
, respectively. In the 1970 – 1984 period, while
forest cover was reduced from 59.5 to 19.7 %, the landscape lost on average 6.8 Mg C
ha
-1
year
-1
; in the 1984-2000 period forest cover was further reduced to 12.6%, but the
landscape lost only 0.39 Mg C ha
-1
year
-1
, as forest loss was partially compensated by an
increase in shade coffee systems. Conversion of all current sun coffee to shade coffee
systems while protecting the remaining forest, could increase average landscape level C
stocks by 10 Mg ha
-1
over a time frame of say 20 years, or 0.5 Mg C ha
-1
year
-1
.
Keywords: allometrics, carbon stock, coffee, soil carbon, wood density
Introduction
Changes in carbon (C) storage in terrestrial ecosystems as a consequence of human land use have
been simplified in the Kyoto protocol to a forest – nonforest dichotomy
1
, and its derivatives
(deforestation, reforestation, afforestation). As most definitions of ‘forest’ depend on a threshold
land cover fraction by woody perennials, derived systems such as coffee plantations with or
without shade trees may fall under the definition. The variation in C stocks within the forest
category, whatever operational definition once chooses, is considerable and most of the changes
due to a gradual degradation or aggradation of C stocks can remain unnoticed if one uses only
two land cover classes. A more refined C accounting system is clearly needed to clarify changes
in terrestrial C storage, and preferably one that includes all land cover types, whether natural or
man-made. Conversion of natural forest to smallholder farms can be dominated by tree crops: in
Indonesia most of the rubber and coffee is produced by smallholders, who generally do not
follow a monocultural plantation system. In fact coffee can be the start of a complex multistrata
agroforest
2
such as the damar (Shorea javanica) gardens of Krui, W. Lampung, where ‘shade
Accepted for the special issue of the ‘Journal of Science in China’ on “Impacts
of land use change on the terrestrial carbon cycle in the Asian Pacific region”

2
trees’ become a dominant feature of the vegetation and major source of income through the resin
that can be harvested from them. In other coffee-based systems fruit trees and/or timber trees
provide income in addition to the coffee and provide farmers with flexibility in maintaining
income despite the boom-or-bust cycles of the world market for coffee.
Apart from the direct value
3
that soil and aboveground C stocks in shade coffee systems
have for the farmer, increased terrestrial C storage is now seen as an ‘environmental service
function’ that delays the increase of the greenhouse effect and the climate change this causes.
The providers of this function, i.c. the managers of land resources, including farmers, are not
currently linked to the beneficiaries, especially the countries or their citizens with the highest C
emissions due to fossil fuel use and the moral and contractual obligation to reduce net emissions.
Mechanisms to provide such a link can take the form of ‘projects’, with a clearly delineated area,
time frame and set of responsibilities, or that of a more generic ‘program’ for a larger area (e.g. a
province or other part of a national administrative system) with accounting of changes in total C
stocks. While the current emphasis still is on ‘projects’, the level of control needed to guard
against ‘leakage’ (negative effects on C stocks elsewhere) and the concerns for ‘additionality’
(only covering costs beyond what would make sense from a private profitability perspective)
lead to high transaction costs and tensions between the outside entities and local administrative
units
4
. In the longer term improved national C accounting systems may allow a more efficient
‘programmatic’ approach to be followed, with C credits derived from actual changes in national
C stocks. The accuracy of IPCC national C inventory schemes in Asia, however, is limited by
uncertainties in the (changes) in areas under various types of land cover, as well as in the
appropriate ‘activity’ data, or typical C stocks to be assigned to the various land cover types
5,6
.
The Alternatives-to-slash-and-burn program has developed a protocol for assessing above
and belowground C stocks of tropical land use systems that has been applied in the three main
tropical continents
7,8,9,10
. The assessment includes trees, using allometric biomass equations,
understorey vegetation, litter layer and soil C stocks in a nested sampling approach, to evaluate
total C stocks in sample plots. Data on sample plots in different stages of the life cycle of a land
use system are then used to derive a ‘time-averaged’ C stock for the land use system, that in
combination with area-based data on land use change can lead to estimates of changes in
terrestrial C stocks for a well-defined area over a specified period of time. Uncertainties in the
estimates originate at the various steps between field level observations in sample plots and the
final total C stock calculation
6
.
Generic equations for mixed tropical forests in different climatic domains
11
may not be
valid in view of differences in wood density between primary and secondary forests
12
.
An
equation
13
,
14
that includes wood density and a parameter from the relationship between tree
height and stem diameter may be more reliable. A method
15
for estimating allometric equations
on the basis of fractal branching rules, uses tree-specific parameters that can be obtained non-
destructively, for root systems (belowground trees) as well as aboveground. For pruned coffee
trees adjustment for the modified diameter-height relationship is probably needed, while
bamboo, palms and bananas require specific equations.
Soil organic matter can play a positive role in crop production and as such the interests of
farmers and external stakeholders in maintaining high soil C stocks can coincide. For the various
coffee systems for acid soils in the humid tropics an initial assessment of the direct benefit of soil
C can focus on its assumed role in mitigating the effects of acidity and aluminium toxicity
16
.
In the research reported here, we set out to quantify the field-level changes in above- and
belowground C stocks of a conversion of forest to ‘sun’ or ‘shade’ coffee systems, the historical

3
landscape level changes in C stocks for a benchmark area and scenarios that may lead to a partial
recovery.
Material and methods
Study area
Measurements were made in the Sumber Jaya subdistrict (104
o
19’ – 104
o
34’ E, 4
o
55’ – 5
o
10’ S;
780 – 1700 m a.s.l.; mean rainfall 2500 mm year
-1
; W.Lampung, Sumatra, Indonesia) that
approximately equals the catchment area of the Way Besai, one of the main contributaries to the
Tulang Bawang river that flows east through the lowland peneplain. The Way Besai flows
almost full circle around the Bukit Rigis (1623 m a.s.l.) in the center of the catchment, with its
top still covered in forest; other forest remnants are left intact on the steeper outer edges of the
catchment area. The major soils are inceptisols (Dystropepts, Dystrandepts and Humitropepts)
with some entisols (Troporthent). The area has a long history of human use, as evidenced by 8-
11
th
Century megalithic remains in the area with Chinese ceramics indicating long distance trade
relations
17
. Human population density was low till about 1900
18
, when the area was settled by
Semendo migrants from S. Sumatra province followed by Javanese and Sundanese in the 1950’s
and 1970’s.
Plot-level C stocks
In a first survey 24 plots were sampled in Sumber Jaya in September 1999, including the
remnant forests on the ridge top, multistrata ‘shade coffee’ gardens (2, 8, 10, 20 and 30 years
after slash-and-burn land clearing), monoculture ‘sun coffee’ (1, 2, 8, 9, 20, 21 years old) and
fallow (1 and 20 years old). In each plot the diameter and height of live and dead trees was
measured in 40 x 5 m
2
plots with litter and understorey samples in subplots (for full protocol
see
10
. A full description of this survey, including botanical composition of the plots will be
published elsewhere
19
. In a second survey the composition of 19 multistrata ‘shade coffee’ plots
was described.
Trees
Tree biomass (W, dry weight) was estimated using the allometric equation of
13
on the basis of
stem diameter at 1.3 m above the ground (D):
W = 0.11 ρ D
2+c
where ρ is the wood density and the coefficient c is based on the allometric relation between tree
height (H) and D: H = a D
c
(default value for c = 0.62).
For comparison the generic allometric equation for humid tropical forests according to
11
was
used: W = 0.118 D
2.53
.
To avoid the need for measuring wood density ρ for every individual tree, a database of
literature values was developed, recording lower bound, upper bound and medium values.
Currently the database holds entries for 2800 tree species and will be shortly made available via
www.icraf.cgiar/sea. Wood density (at a standard 15% moisture content) can be classified as
light (density less then 0.6 Mg m
-3
), medium (between 0.6 to 0.75 Mg m
-3
), heavy (0.75 to 0.9
Mg m
-3
) and very heavy (more than 0.9 Mg m
-3
).
For pruned coffee, bamboo and banana separate allometric equations were used, derived
for similar conditions in East Java (Fig. 1) as quoted in
10
. For pruned coffee we used W =
0.281D
2.06
, the power of which agrees with
13
equation, as c = 0.08 (H = 1.79 D
0.08
). For banana
biomass based on pseudostem diameter we used: W = 0.030 D
2.13
. For bamboo we used: 0.131
D
2.28
. Total C content was calculated from biomass assuming a C content (per unit dry weight) of
0.45.

4
Figure 1. Allometric relationships between (pseudo)stem diameter and plant height or biomass
(dry weight) for pruned coffee
10
Roots
Fractal branching models
15
provide a transparent scheme for deriving tree-specific scaling rules
on the basis of easily observable, non-destructive methods. These models repeatedly apply the
Pruned coffee
024681012
Tree height, m
1.8
1.9
2.0
2.1
2.2
024681012
Tree biomass, kg/tree
0
5
10
15
20
25
30
Banana
0 5 10 15 20 25 30
Tree height, m
0
2
4
6
8
10
0 5 10 15 20 25 30
Tree biomass, kg/tree
0
10
20
30
40
50
Bamboo
Stem diameter at 1.35 m, cm
02468
Tree height, m
0
2
4
6
8
10
12
Stem diameter at 1.35 m, cm
02468
Tree biomass, kg/tree
0
2
4
6
8
10
12
14
y = 1.79 x
0.0797
R
2
= 0.84
y = 0.2811 x
2.0635
R
2
= 0.95
y = 0.707 x
0.6835
R
2
= 0.81
y = 0.030 x
2.13
R
2
= 0.99
y = 1.45 x
0.96
R
2
= 0.94
y = 0.131 x
2.278
R
2
= 0.95

5
same rules (equations) to derive subsequent orders of the branching process. When a rule is
added for stopping when a certain minimum size is reached, the total size of a branched root (or
aboveground tree) can be derived on the basis of initial (proximal) root (stem) diameter. In a
spreadsheet model (functional branch analysis, or FBA) available through
www.cgiar.org/icraf/sea/agromodels/wanulcas/wanulcas.htm, the relations between five input
parameters and the parameters of the allometric biomass equation a and b can be explored. Five
parameters describe the branching process: n = average number of branch offspring per
branching point, p = proportionality of stem cross sectional area before and after branching, q =
allocation of offspring cross sectional area to the largest link, L
m
= length of a link (internode) at
minimum stem diameter and r = slope of the regression of link length on stem diameter.
Proximal roots were exposed on 3 coffee trees each of age 1, 3, 7 and 10 years,
respectively and their diameter and orientation was recorded. A number of roots of 6 coffee
plants was exposed and the parameters n, p, q, L
m
and r were derived for a total of 34 branching
points. With the FBA model
15
, the following allometric equation was derived for coffee roots: W
= 0.0074 D
prox
3.14
.
Soil Carbon
For the 24 samples in the first land cover survey soil samples of the 0-5, 5-15 and 15-30 cm
depth layer were collected and analyzed for Corg (Walkey and Black), texture (sand, silt clay),
pH (1M KCl), exchangeable bases (Na, K, Ca and Mg), exchangeable Al and acidity.
A dimensionless 'C saturation deficit', C
satdef
, was calculated as the difference between the
current C
org
content and a reference content, C
org, ref
which is supposed to indicate the undisturbed
forest condition.
C
satdef
= (C
org, ref
- C
org
) / C
org, ref
= 1 - ( C
org
/ C
org, ref
)
The ratio
20
of measured C
org
and a reference C
org
value for forest (top) soils of the same texture
and pH can serve as 'sustainability indicator'. The current version incorporates a generic C
distribution with depth
10
. The equation for C
org,,ref
for Sumatra is:
C
ref(adjusted)
= (Z
sample
/ 7.5)
-0.42
exp(1.333 + 0.00994 * %Clay + 0.00699 * %Silt – 0.156 * pH
KCl
+
0.000427 * Elevation + 0.834 (if soil is Andisol) + 0.363 (for swamp forest on wetland soils)
The depth adjustment factor for the layers 0-5, 5-15 and 15-30 is 1.59, 0.89 and 0.63,
respectively.
Time-averaged C stock
For rotational land use systems, which include a clear felling at the start (or end) of every
production cycle, a ‘time-averaged’ C stock can be used for landscape scale assessments, as it
reflects a ‘spatially averaged’ value on the assumption of a steady rate of renewal
8
. For the
coffee production systems in Sumber Jaya, we can distinguish an establishment period of
approximately 5 years from a mature phase in which continuous tree rejuvenation (by direct
grafting of the coffee and gradual replacement of companion trees) causes fluctuations around an
average value, without a clear felling phase.
Landscape-level change in C-stock
Local land use maps
21
from 1970, 1978, 1984 and 1990 were complemented by June-2000
Landsat ETM imagery. For each of these years we calculated landscape-level C stock by
multiplying areas per land use with the time-averaged C stock of the land use type.

6
Figure 2. Cumulative frequency of wood density estimates found in a broad-based literature
survey, A. for a species list from the secondary forest in Sumber Jaya, B. for a species list of the
multistrata coffee gardens in Sumber Jaya; these data are extracted from a wood density data
base that currently includes 3500 tree species, with a lower- and upper bound estimate for most
species, as well as a midpoint estimate for all; the database will be made available via ICRAF-
SEA web-site
Results and Discussion
Biomass carbon
The median mid-range wood density for tree species found in the remnant and secondary forest
was 0.61 Mg m
-3
(with a distribution of species as follows: light 40.8%, medium 31.7%, heavy
15% and very heavy 12.5%) (Fig. 2). The median mid-range wood density for tree species found
in multistrata coffeee gardens was 0.75 Mg m
-3
, (light 39.2%, medium 38.5%, heavy 15.2% and
very heavy 6.5%).
Compared to the ‘generic’
11
allometric equation, use of tree-specific allometrics that
include estimates of wood density tend to lead to lower biomass estimates (Fig. 3), especially in
the low-to-medium biomasss categories. As all earlier ASB results
8,9
used the generic
11
equation,
some caution is needed when comparing the results.
Three remnant/secondary forests were included in the survey, with a mean estimated
aboveground biomass of 390 Mg ha
-1
. Only 8% of this amount derived from the litter, dead trees
and undergrowth. For two fallow plots of 1 and 20 year after the last slash-and-burn land
clearing, aboveground bio- & necromass was estimated to be 9 Mg ha
-1
and 170 Mg ha
-1
,
respectively. The younger fallow did not contain woody biomass; for the older plot 10% of the
total was derived from litter, dead trees and undergrowth. The total biomass average in shade
coffee gardens from 2 years to 30 years old was around 92 Mg ha
-1
; 25% of this value derived
from litter, dead trees and undergrowth. The total biomass average in monocultural sun coffee
A. Secondary forest
0.00.20.40.60.81.0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
B. Multistrata coffee
0.0 0.2 0.4 0.6 0.8 1.0
low bound
midpoint
upper bound
Cumulative frequency Cumulative frequency
Wood density, Mg m
-3
Median = 0.61
Median = 0.75

7
Figure 3. Comparison of aboveground biomass estimates based on the (generic tropical forest)
11
allometric equation and estimates based on the equation
13
that accounts for the wood density of
the tree species (using the mid range estimates from a broad literature data review)
between 1 and 21 years old was 44 Mg ha
-1
, 48% of which was litter, dead trees and
undergrowth. The difference between shade and sun coffee was thus close to 50 Mg ha
-1
of
biomasss, or 20 Mg ha
-1
of aboveground C stock.
Figure 4. Relationship between aboveground C stock and age of coffee gardens (presumably
since last clear-felling or slash-and-burn land clearing, but note that for some of the shade coffee
gardens the age of the coffee plants is used as X-axis as the age of the garden is unknown)
In the second survey, the average C stock in aboveground biomass of coffee trees in
shade coffee gardens between 6 and 40 years of age was 18.4 Mg ha
-1
, with a standard deviation
of 4.0, and the average for non-coffee trees was 29.6 Mg ha
-1
with standard deviation of 18.9.
By plotting the total aboveground C stock data against age of the garden (Fig. 4, note some
confusion between age of garden and age of coffee, indicating that not all coffee was planted
after clear-felling), we can estimate the annual C stock increments as roughly 1 and 2 Mg C ha
-1
0
100
200
300
400
500
0 100 200 300 400 500
Estimate based on Ketterings (2001), Mg/ha
Estimate based on Brown
(1977), Mg/ha
line 2:1
line 1:1
y
=
0
.
9
7
x
y
=
1
.
8
6
x
y
=
3
.
4
4
x
0
50
100
150
200
0 102030405060708090100
Age of coffee garden, year
Carbon biomass, Mg ha
-1
mixed
mixed (incl. older trees)
monoculture
fallow
forest
damar

8
Figure 5. Estimates of above and belowground biomass of coffee and shoot:root ratio, based on
allometric equations for stem diameter (empirical equation) and proximal roots (derived from
fractal branching model with parameters derived from coffee in Sumber Jaya)
year
-1
for sun and shade coffee respectively. This is less than the inferred C accumulation rate for
bush fallow here (although we do not have enough data points), or in previous data sets for
Sumatra
8
. Three sample points for the damar agroforest of Krui (derived from data in
2
fit
remarkably well to the extrapolated line for ‘shade coffee’.
While shoot:root biomass ratios for tropical forests under normal upland conditions are
normally in the neighbourhood of 4:1, our data on coffee suggest a decrease of the shoot:root
ratio with age to about 2:1, for the regularly shoot-pruned coffee trees in Sumber Jaya (Fig. 5).
Soil C
Soil organic matter contents for the forest sites are in the range expected for Sumatra
20
, once the
soil texture and soil pH are taken into account in the Cref value and adjustments were made for
sampling depth. The relative soil C content (C/C
refadj
) was around 0.79 (0.82 for 0-5 cm, 0.84 for
5-15 and 0.71 for the 15-30 cm layer) in the forest sites, around 0.60 for the mixed coffee
gardens (on average 0.58, 0.70 and 0.52 for the three soil layers, respectively) and 0.45 for
monoculture coffee (0.43, 0.52 and 0.41, respectively). When plotted against the time since
forest conversion (Fig. 6), there is an indication of decline of the C/Cref value with time in the
coffee monocultures, and a possible recovery towards forest values in the mixed coffee gardens.
Time-averaged C stocks and landscape level changes
The data for soil C, root biomass and aboveground biomass of trees, necromass, litter layer and
understorey or herb layer, were combined to derive time-averaged C stocks (above a soil depth
of 0.3 m) for the first 25 years of sun and shade coffee of 52 and 82 Mg C ha
-1
, that are
considerably below that of the remnant forest (262 Mg C ha
-1
) or the young secondary forest
(remnant of ‘shifting cultivation’), at 96 Mg C ha
-1
.
Coffee age, years
024681012
Estimated biomass (kg/tree)
0
5
10
15
20
25
Shoot/root ratio
0
1
2
3
4
5
6
Root
Shoot
Shoot/root ratio

9
Figure 6. Relationship between the C
org
content of the soil relative to the reference
value (C
refadj
) as a function of age, for the
open coffee monocultures and the mixed
coffee systems, that include other trees;
linear regression lines with 95% confidence
intervals are drawn; for comparison three
samples of damar agroforest
2
are added
By multiplying the land area under the various land cover categories with the current
estimates of time-averaged C stocks, we obtain estimates of landscape-level changes in C stocks
(Fig. 7). From a level of about 200 Mg C ha
-1
in 1970 (59.5 % forest cover) the decrease was 6.8
Mg C ha
-1
year
-1
in the first 14 years to a value of 92 Mg C ha
-1
in 1984 (19.7% forest cover).
Figure 7. Landscape level C stocks above a soil depth of 0.3 m for Sumber Jaya in the period
1970 – 2000, with an extrapolation to the future assuming all coffee gardens to become ‘shade
coffee’, while the remaining forest is left intact
C/Cref
0
1
2
Age since defor vs Forest+Fallow 0
Age since defor vs Coffee mono 0-5
Plot 2 Regr
Plot 2 Conf1
Age since defor vs Coffee mix 0-5
Plot 3 Regr
Plot 3 Conf1
C/Cref
0
1
2
Age, year
0 5 10 15 20 25 30 35
C/Cref
0
1
2
0 - 5 cm depth
5 - 15 cm depth
15 - 30 cm depth
Forest+Fallow
Coffee mono
Coffee mixed
0
50
100
150
200
1970 1980 1990 2000 2010
Time, year
Total C stock, Mg/ha
0
20
40
60
80
100
1970 1980 1990 2000
Time, year
Percentage land cover
Upland
crops &
fallow
Sun
coffee
Shade
coffee
Paddy rice
Settlements & ponds
Forest
Forest
Other
Shade coffee
2025?
maximum
change
scenario

10
The further loss of forest cover in the period 1984 – 2000 (till 12.6%) lead to a landscape
level loss of only 0.4 4.1 Mg C ha
-1
year
-1
to 86 Mg C ha
-1
in 2000, as shade coffee systems
replaced some of the sun coffee, while grassland as land cover category was reduced from 15.6%
in 1984 to 9.9% in 2000.
Using the same data we can explore a scenario where all currently open land and sun
coffee systems are converted to shade coffee over the next, say, 25 years, while current forest
cover is maintained. This may lead to a landscape level C stock of 102 Mg C ha
-1
– which is 10
Mg C ha
-1
above the 1990 value.
Concluding remarks
The differences in C stocks between sun and shade coffee (about 30 Mg ha
-1
in time-averaged C
content) may appear relevant when valued at a current global C market price
4
of 8 $ Mg
-1
and a
(discounted, private prices) net present value (NPV) of the coffee production systems
22
of 450 –
3250 $ ha
-1
.The lowest NPV reflects pioneer coffee monocultures with insecure land tenure and
low management intensity, the highest ones complex coffee gardens with secure tenure and
medium management intensity. These results were based on 1997 estimates of long term price
averages, before the ‘monetary crisis’; at current prices the NPV might be only a third of this
amount). Yet, the chances that smallholder coffee growers in Sumber Jaya can directly capitalize
on such a world market value for additionally stored C are slim. First of all, transaction costs for
any marketing and certification may well exceed or come close to the supposed market value,
unless substantial economies of scale can be achieved by grouping of smallholder producers
4
.
Secondly, current data suggest that shade coffee systems, that maintain higher C stocks, are
already more profitable than low-C stock sun coffee systems, so a transition from sun-to-shade
coffee would not pass the ‘additionality’ criterion. As the ‘leakage’ aspect for any rewards on
coffee systems that have recently been obtained by forest conversion is daunting, the only chance
may be for larger units, for example the Sumber Jaya subdistrict of 700 km
2
, to become the scale
of analysis and C accounting. In that case an existing local administrative unit would become
responsible for at least maintaining current landscape-level C stocks, with an option for
increasing it with the said 10 Mg C ha
-1
in the next decade(s), if overall incentives (e.g. tenurial
& tax) favour shade-coffee development. A further point to note is that although shade coffee
systems as a category represent a ‘win win’ solution for both private and external
(environmental) interest groups when compared to ‘sun coffee’, there may very well be negative
trade-offs between C stocks and profitability within the ‘shade coffee’ class of land use systems.
A more detailed classification system would then be needed, along with more refined techniques
for monitoring land use by remote sensing data. For the time being, the chances of C
sequestration reward mechanisms are larger for smallholders in more C-depleted landscapes, as
for example in the ASB N. Lampung benchmark area in the lower reaches of the Tulang Bawang
river where deforestation has been completed and smallholders now start to re-plant trees in
depleted landscapes
23,24
.
References
1. Watson, R. T., Noble, I. R., Bollin, B., Ravindranath, N. H., Verado, D. J. and Dokken, D. J., Land
Use, Land-Use Change and Forestry. A Special Report of the IPCC. Cambridge University Press,
Cambridge, UK, 2000, 377
2. Vincent, G., de Foresta, H. and Mulia, R., Predictors of tree growth in a Dipterocarp based
agroforest: a critical assessment. Forest Ecology and Management, 2002 (in press)
3. Lyngbæk, A. E., Muschler, R. G. Sinclair, F. L., Productivity and profitability of multistrata organic
versus conventional coffee farms in Costa Rica Agroforestry Systems, 2001, 53: 205-213

11
4. Tomich, T. P., de Foresta, H., Dennis, R., Ketterings, Q. M., Murdiyarso, D., Palm, C. A., Stolle, F.,
Suyanto, S. and Van Noordwijk, M., Carbon offsets for conservation and development in Indonesia?
American Journal of Alternative Agriculture, 2002 (in press)
5. Magcale-Macandog, D. B., Status of GHG inventory for the LUCF sector in the Philippines,
Indonesia and Thailand. Proceedings of the IGES/NIES workshop on GHG inventories for Asia-
Pacific region, 9-10 March 2000 at Shonan Village, Hayama, Japan. Institute for Global
Environmental Strategies. 1560-39 Kamiyamaguchi, Hayama, Kanagawa, 2000, 110-137.
http://www.iges.or.jp/cc/napiid/PDF/2_LULUCF1.pdf
6. van Noordwijk, M., Hairiah, K. and Sitompul, S. M., Reducing uncertainties in the assessment at
national scale of C stock impacts of land use change. In: D.B. Magcale-Macandog (Ed.) Proceedings
of the IGES/NIES Workshop on GHG Inventories for Asia-Pacific Region. Institute for Global
Environmental Strategies (IGES), Hayama, Japan, 2000, 151-163.
http://www.iges.or.jp/cc/napiid/PDF/2_LULUCF2.pdf
7. Woomer, P. L. and Palm, C. A., An approach to estimating system carbon stocks in tropical forests
and associated land-uses. Commonwealth Forestry Review, 1998, 77: 181-190.
8. Palm, C. A., Woomer, P. L., Alegre, J., Arevalo, L., Castilla, C., Cordeiro, D. G., Feigl, B., Hairiah,
K., Kotto-Same, J., Mendes, A., Moukam, A., Murdiyarso, D., Njomgang, R., Parton, W. J., Ricse,
A., Rodrigues, V., Sitompul, S. M. and van Noordwijk, M., Carbon sequestration and trace gas
emissions in slash-and-burn and alternative land uses in the humid tropics. ASB Climate Change
Working Group Final Report, Phase II. ICRAF, Nairobi, 1999, 36.
9. Woomer, P. L., Palm, C. A., Alegre, J., Castilla, C., Cordeiro, D. G., Hairiah, K., Kotto-Same, J.,
Moukam, A., Ricse, A., Rodrigues, V. and van Noordwijk, M., Slash-and-Burn Effects on Carbon
Stocks in the Humid Tropics. In: R Lal, JM Kimble and BA Stewart (eds.) Global Climate Change
and Tropical Ecosystems. Advances in Soil Science. CRC Press, Inc. Boca Raton, FL. USA, 2000,
99-115.
10. Hairiah, K., Sitompul, S. M., van Noordwijk, M. and Palm, C. A., Methods for sampling carbon
stocks above and below ground. ASB LectureNote 4B, International Centre for Research in
Agroforestry (ICRAF), Bogor, 2001. http://www.icraf.cgiar.org/sea/Training/Materials/ASB-
TM/Datas/ASBLecNote%204B.pdf
11. Brown, S., Estimating biomass change of tropical forest, a primer. FAO Forestry Paper 134, FAO,
Rome, 1997.
12. Fearnside, P. M., Wood density for estimating forest biomass in Brazilian Amazonia. Forest Ecology
and Management, 1997, 90(1): 59-87.
13. Ketterings, Q. M., Coe, R., van Noordwijk, M., Ambagau, Y. and Palm, C. A., Reducing uncertainty
in the use of allometric biomass equations for predicting above-ground tree biomass in mixed
secondary forests. Forest Ecology and Management, 2001, 120: 199-209.
14. Brown, I. F., Martinelli, L. A., Thomas, W. W., Moreira, M. Z., Ferreira, C. A. C. and Victoria, R. A.,
Uncertainty in the biomass of Amazonian forests: an example from Rondonia, Brazil. Forest Ecology
and Management, 1995, 75: 175-189.
15. van Noordwijk, M. and Mulia, R., Functional branch analysis as tool for scaling above- and
belowground trees for their additive and non-additive properties. Ecological Modelling, 2002, 149(1-
2): 41-51.
16. Wrigley, G., Coffee. Longman, Singapore, 1988.
17. Edward McKinnon, E., A note on finds of early chinese ceramics associated with megalithic remains
in Northwest Lampung. J. Southeast Asian Studies, 1993, 24: 227-238.
18. Benoit, D., Pain, M., Levang, P. and Sevin, O., Transmigration en spontanees en Indonesie/
Transmigration and spontaneous migrations in Indonesia: Lampung Province. ORSTOM &
Transmigration Department, Jakarta, 1989.
19. Gillison, A. N., Liswanti, N., Budidarsono, S., van Noordwijk M., and Tomich T. P., Impact of
cropping methods on biodiversity in coffee agroecosystems: Sumatra, Indonesia. (submitted)
20. van Noordwijk, M., Cerri, C., Woomer, P. L., Nugroho, K. and Bernoux, M., Soil carbon dynamics in
the humid tropical forest zone. Geoderma, 1997, 79: 187-225.

12
21. Syam, T., Nishide, H., Salam, A. K., Utomo, M., Mahi, A. K., Lumbanraja, J., Nugroho, S. G. and
Kimura, M., Land use and cover changes in a hilly area of South Sumatra, Indonesia (from 1970 to
1990). Soil Science and Plant Nutrition, 1997, 43: 587-599.
22. Budidarsono, S., Kuncoro, S. A., Tomich, T. P., A profitability assessment of robusta coffee systems
in Sumberjaya watershed, Lampung, Sumatra, Indonesia. Southeast Asia Policy Research Working
Paper No. 16, ICRAF-SEA, Bogor, Indonesia, 2001.
23. van Noordwijk, M., Murdiyarso, D., Hairiah, K., Wasrin, U.R., Rachman, A. and Tomich, T. P.,
Forest soils under alternatives to slash-and-burn agriculture in Sumatra, Indonesia. In: Schulte A and
D Ruhiyat (eds.) Soils of Tropical Forest Ecosystems: Characteristics, Ecology and Management.
Springer-Verlag, Berlin, 1998,175-185.
24. Roshetko, J. M., Delaney, M., Hairiah, K., and Purnomosidhi, P. Carbon stocks in Indonesian
homegarden systems: Can smallholder systems be targeted for increased carbon storage? American
Journal of Alternative Agriculture, 2002. (in press).