Strategic Information On Changes In Carbon Stocks And Land-Use

“Best bet” Land-use Systems

Thematic reports

Carbon Sequestration And Trace Gas Emissions

 

Unique id: IDAOJYZB

Source file: D:\Projects\ASB\ASB Country and Thematic reports\Climate Change WG Report\phase2final999.xml

 

Authors: C. A. Palm, P. L. Woomer, J. Alegre, L. Arevalo, C. Castilla, D. G. Cordeiro, B. Feigl, K. Hairiah, J. Kotto-Same, R. Lasco, , A. Mendes, A. Moukam, D. Murdiyarso, R. Njomgang, W. J. Parton, A. Ricse, V. Rodrigues, S. M. Sitompul, M. van Noordwijk

 

--------------------------------------------------------------------------

The role of tropical forests in the global carbon (C) cycle has been debated over the past 20 years, with several estimates of its contribution to the increase in atmospheric carbon dioxide (Houghton et al., 1987; Detwiler and Hall, 1988).  Today there is general agreement, based on land-use change data and atmospheric data, that the tropics are a net source of C to the atmosphere, in the range of 1.1 to 2.1 Pg C y-1  (Houghton, 1997).  The primary cause of this net source is deforestation in the tropical zone, with Asia and Latin America accounting for over 80% of the flux (Houghton, 1997). The net CO2 flux is a result of land-use change and depends on how quickly land uses are converted, the biomass of  the vegetation that is cleared, the fate of the carbon cleared, the biomass of new vegetation, the time course of the subsequent land-use systems, and the regrowth rates of vegetation.  Much of the uncertainty in the values of CO2 flux from the tropics is a result of inadequate estimates for these parameters (Houghton, 1997). In particular, there is little information on the carbon sequestration potential of many of the land-use systems of the humid tropics (Houghton et al., 1993).

 

One activity of the ASB project was to characterize the patterns of land clearing and subsequent land use at the benchmark sites and to quantify the changes in carbon stocks associated with land clearing and establishment of different land-use systems.  Standardized methods were established to measure carbon stocks in the forests, the various land-use systems established following slash and burn clearing, and promising “best-bet” alternatives at the different sites (Woomer and Palm, 1994, 1998).  These data were used to calculate both the immediate and longer term losses of carbon associated with slash-and-burn clearing and to identify those land-use systems or “best-bet” alternatives that sequester the most carbon.

 

Carbon stocks were measured in the soils and vegetation in 94 sites in the three ASB benchmark countries (Brazil, Cameroon, Indonesia) and in an additional 22 sites in Perú.  The sites sampled in each country included undisturbed or selectively logged forests as the reference point; areas that had been recently slashed, burned, and cropped; and areas that were subsequently planted to pastures, tree plantations or agroforests, or areas abandoned to fallow regrowth.  Details of the sites sampled are provided in Appendix 1. This dataset, compiled by the ASB benchmark team, is unique in that it provides data collected and analyzed by standardized methods across sites.  In addition, the information in this dataset on the carbon stocks and carbon accumulation rates in young fallow vegetation and agroforestry and plantation systems are rare for the tropics.

 

A paper synthesizing the data on C stocks across sites was presented at an international meeting in Brazil, Carbon Pools and Dynamics in Tropical Ecosystems.  This paper (Woomer et al., 1999) summarized the above-and below-ground carbon stocks for the different land-use types.  This analysis enables calculations of carbon losses when one land-use type is converted to another.  A global summary of the total system carbon (TSC) in a 20 year "traditional" slash-and-burn sequence is  (t ha-1):original forest (305) to burned cropland (52) to bush fallow (85) to tree fallow (136) to secondary forest (218).  Logging reduced forest system carbon by 124 t ha-1.  Ten year-old pastures in Brazil and 13 year-old Imperata spp. grasslands in Indonesia contained less TSC than croplands (-4.8 and -5.3 t C ha-1, respectively).  Recently established agroforestry systems contained more TSC than did croplands (+13.2 t C ha-1).  Mature agroforests (130 t C ha-1) contained significantly greater TSC than croplands, pastures

and grasslands but significantly less than secondary forests of similar age. Carbon sequestration rates for natural fallows were 8.4 t C ha-1 yr-1 following land abandonment.  Agroforestry systems, if established at the time of initial land clearing, sequestered 3.3 t C ha-1 yr-1. Pastures/grasslands tended to lose C at a slow rate (240 kg ha-1 yr-1).  Land-use systems where trees were planted and managed had greater potential to sequester C than did field crops or pastures, but with sequestration rates less than those of natural succession.

 

1. 1 Above-Ground Time-Averaged Carbon Stocks

While the global synthesis presented above provides us with information on the trends in C stock changes with differing land-use conversions following slash and burn agriculture, it does not necessarily provide a means of assessing which land-use systems sequester more carbon. To compare the carbon sequestration potential of a land-use system, it is necessary to know the average C stored in that land-use system over the rotation time of that system (LUSCta).  In other words, it is not necessarily the maximum C stock of the systems that is important for considering net C fluxes, but rather the average C stock of the system over time.

 

A forest system has a fairly constant C stock, whereas clearing the forest and establishing tree plantation results in an initial large loss followed by gradual accumulation of C.  The plantation may eventually reach 50 to 80% of the C stock of the forest, but the time to accumulate that stock will vary by tree species, management, soils, and climate.  The time- averaged C stock depends on the carbon accumulation rates,  the maximum C stored in the system, the time it takes to reach maximum C, and the rotation time of the system (Figure 1). The methodology for determining time-averaged C stocks and an example are presented in Appendix 2.  The time-averaged carbon stocks, or carbon sequestration potential, were calculated for the above-ground component of the different slash and burn and alternative land-use systems for the three benchmark sites.

 

Benchmark results 

The average maximum carbon stock, regrowth rates, and time-averaged carbon, along with the standard deviation or range in calculations, are summarized by land-use system and benchmark country in Table 1.  Detailed worksheets for each country are included in Appendix 3.  The results on carbon stock measurements and time-averaged C calculations are presented and discussed based on average values.  In addition, Table 1 and Appendix 3 include the standard deviation of the measured stocks and range in calculated time-averaged C to provide an indication of the precision of the estimates.

 

BrazilThe average carbon stock (the above-ground vegetation plus litter) of the 4 selectively logged forests in Brazil was approximately 150 t C ha-1, ranging from 130 to 175.  The average value, 150 t C ha-1, was compared to the maximum C sequestered (Cmax) and time-averaged C (LUSCta) in the following land-use systems (Table 1, Figure 2a):

                   pastures, both traditional and improved;

                  monoculture coffee plantations (1000 plants ha-1), assuming a 7 year establishment phase plus 5 more years of production for a total rotation time of 12 years;

                  multistrata agroforestry systems (includes 3 systems - coffee + rubber, coffee + banderra, cupuacu + pupunha + castanha) with an establishment phase of 12 years and rotation time of 20 years;

                  annual crop-fallow cycles with  3 years of cropping and 5 years of natural fallow;

                  annual crop-improved tree fallow (inga or senna) cycles with 3 years of cropping and 5 years of fallow.

 

The traditional land-use practice of converting forestlands to pastures results in only 2% of the above-ground C of the forest.  The average rotation time of a pasture is 8 to 10 years before re-establishment, but the rotation time, does not have much effect on C storage in pastures because of the constant biomass maintained through grazing.  Improving pastures either through management or planting legumes does not increase the carbon storage or time-averaged C stocks above that of the traditional pastures.  Lands planted to perennial tree crops attain a maximum C stock of as little as 15 t C ha-1 (10% of forest C stock) for monoculture coffee to as much as 90 t C ha-1 (54% of forest) for multistrata agroforestry systems of 20 year rotation times.  The time-averaged C stocks, however, are only 5 to 26% that of the forest.  For land put into an annual crop-fallow rotation, the maximum C stock of a natural fallow of 5 years is 20 t C ha-1, compared to 34 t C ha-1 for an improved tree fallow, which is 23% of the forest.  The time-averaged C stock of the crop-5 year natural fallow is only 6.86 t C ha-1 (5% of the forest). The value increases to only 11.5 t C ha-1 for improved tree fallows. The slight increase in C storage and time-averaged C with the improved fallow is due to their high C accumulation rate of 6.86 t C ha-1 y-1, compared to 3.91 t C ha-1 y-1 for the natural fallow.  The regrowth rates of the natural fallows are within the range, but at the upper end, of other studies in Brazil (Fearnside and Guimaraes, 1996). The C accumulations rates of the multistrata agroforestry systems were high and similar to the improved fallow.

 

Cameroon  The C stocks (above-ground vegetation and litter) of the 6 selectively logged forests sampled in Cameroon averaged about 228 t C ha-1 , ranging from 193 to 252 t C ha-1. This value was compared to data for the following land-use systems (Table 1, Figure 2b).

                  Annual cropping phase of 2 years followed by 4 years of chromolaena fallow,

                  2 years cropping followed by either 9 or 23 years of bush-tree fallow,

                  2 years cropping followed by establishment of cacao (jungle cacao) over 25 years,

                  a 40 year rotation versus a non-rotational cacao system established through gap and understorey plantings of cacao,

                  one year cropping followed by establishment of an oil palm plantation with 146 trees ha-1 with a 7 year establishment phase and a 25 year rotation.

 

The maximum C stock attained in the various crop-fallow systems was 167 t C ha-1, for the traditional long fallow.  The amount is reduced by more than half, to 76 t C ha-1, if the fallow is shortened to 11 years, and further reduced to 12 t C ha-1 with the 4 year chromalaena fallow. The time-averaged C stocks of these crop-fallow rotations are 77, 32, and 5 t C ha-1, respectively.  A mature jungle cacao stand contains about 43% of the C of the forest, ranging from 54 to 131 t C ha-1, with an average of 89 t C ha-1.  If the jungle cacao system is established simply by clearing the understorey and planting cacao, then the time-averaged carbon of this non-rotational system is the same as the carbon stocks measured.  If the system is established through slash-and-burn clearing and cropping followed by planting of cacao with a 25 year establishment phase and total rotation time of 40 years, the time-averaged carbon is 61 t C ha-1. The maximum C and time-averaged C of an oil palm plantation with a 7 year establishment phase and rotation time of 25 years are about half that of the cacao system.

 

The rates of C accumulation (sequestration rates) varied with age of the fallow; beginning with 2.89 t C ha-1 the first two years when chromolaena dominated, increasing to 8.5 t C ha-1 for the next 6 to 10 years.  The overall accumulation rate during the traditional long shifting cultivation fallows was 7.26 t C ha-1.  The C accumulation rate of the rotational jungle cacao was only half that of the natural fallow systems, whereas that of the oil palm plantation was similar at 6.03 t C ha-1.  The rates of C accumulation are quite high compared to most report for the humid tropics but they do fall within the range measured by Szott et al., (1994).

 

Indonesia  The C stocks (above-ground vegetation and litter) of the forests sampled in Indonesia averaged 306 t C ha-1 (376 and 236 t C ha-1) for the two primary rainforest sites and 93 t C ha-1 (49 to 144 t C ha-1) for the logged-over forests.  These forest C stock values were compared to those of the following land-use systems.

                  Annual cropping of 2 years followed by establishment of a rubber plantation (jungle rubber) with a 25 year establishment phase and 30 year rotation time,

                  A non-rotational rubber system established through understorey gap plantings,

                  Establishment of an industrial oil palm plantation with 120 trees ha-1 and an establishment phase of  7 years and rotation time of 25 years,

                  Establishment of an industrial timber plantation of fast-growing trees (Paraserianthes, Eucalyptus, Acacia) with a rotation time of 8 years,

                  An annual cropping system of 7 years of cassava followed by 3 years of imperata.

 

The carbon stock of the logged-over forests in Indonesia is only 24% that of the undisturbed rainforest.  Most land conversion in Indonesia follows after logging.  The maximum C stock of the jungle rubber system is 89 t C ha-1.  The time-averaged carbon for the non-rotational jungle rubber is the same as the average stock (89 t C ha-1), whereas that of the rotational jungle rubber system is only about half, at 46 t C ha-1. The maximum C stored in the monoculture oil palm plantation is about 30% less than that of the rotational jungle rubber (63 t C ha-1), but because of the faster growth and establishment of the oil palm trees, the time-averaged C stock is slightly higher than that of the jungle rubber.  The fast-growing timber plantations likewise reach a similar maximum C stock (74 t C) to that of the jungle rubber but in only 8 years as compared to 25 years.  The shorter rotation time, 8 years compared to 25 and 30 for the oil palm and jungle rubber, results in a slightly lower time-averaged C.  The crop-fallow and cassava-imperata systems have time-averaged C of only 6 and 2 t C ha-1, respectively, less than 2% that of the forest.

 

The C accumulation rates of the jungle rubber systems in Indonesia (3.6 t C ha-1) were similar to those of the cacao systems in Cameroon.  The more intensive plantation systems had much higher C accumulations rates, almost reaching 10 t C ha-1 in oil pulpwood plantations. 

 

1. 2 Below-ground carbon stocks and changes

The preceding comparison includes only the above-ground carbon stocks, because the root and soil data were extremely variable.  The root data in particular were not useful in making comparisons between land-use systems.  Apparently the excavation method used did not adequately sample large roots, so the values for roots in forests and other tree-based systems were underestimates.  These data are not included in the report and will not be discussed.  The soil data were also variable, partially because of textural differences in the soils of the chronosequence sampled at each site, despite attempts to sample similar soils.  Differences in soil C measured in two different land-use systems can in fact be a result of soil textural differences, rather than any effect of land-use.  In order to account for the variability caused by differences in soil texture within a site, the soil C data were normalized using the equation developed by van Noordwijk et al., (1997) for estimating the soil C equilibrium values.  The equation calculates what the equilibrium soil C would be in a natural, undisturbed system. 

 

Calculated forest soil C = Cref = exp {1.333 + 0.00994*%clay + 0.00699*%silt 0.156*pHKCl}.

 

The calculated reference values for each site sampled were then compared to the actual C measured (Cact), to give a relative C value (Crel) = Cact/ Cref.  The Crel values obtained for the forest sites were not always predicted by the equation, so the Crel of each land use was divided by the Crel of the forest within each site, to serve as an estimate of the % change in soil C from a particular land-use transition.  The relative C values of the different land-use systems compared to that of the forest for each of the benchmark sites are presented in Table 2.

 

The data in Table 2 indicate that for all the land-use systems considered in Cameroon there is little or no change in soil C.  Although these data were not corrected for changes in bulk density, this apparent lack of change is consistent with the relatively low land-use intensity in this benchmark area.  Even the cropping systems measured show little change, because they are cropped for only one year prior to abandonment to fallow regrowth.  In contrast to Cameroon, in the more intensive pastures and croplands in Brazil and degraded grasslands and continuous cropping in Indonesia, soil C losses of 11 to 53% were found from conversion of forests.  This wide range in soil C losses depended on the length of time the land had been in the particular land use, the soil type, and topsoil erosion.  In general, the tree-based plantations and agroforestry systems lose less than 20% of the topsoil C, and the complex rubber and cacao agroforests had similar levels of soil C to that of the forests. The relative soil C losses reported for the ASB sites are similar to those reported by Detwiler (1986) in a review of soil C changes with land-use change in the humid tropics.

 

The CENTURY model has also been used to estimate the equilibrium values of soil C for forest soils (a reference soil C value) in Sumatra and to simulate soil C with different land-use conversions (Sitompul, unpublished data). Overall, the impacts of land-use change on soil C pools as predicted by the CENTURY model agree in magnitude and relation with available soil data from Sumatra. The CENTURY model allows for finer distinctions between various land-use scenarios than does the database (which, for example, lumps all perennial crop plantations into a single category and does not allow for distinctions between oil palm, rubber and Paraserianthes plantations, except through their impact on soil pH).

 

1. 3 Carbon losses and potential for carbon sequestration with alternatives to slash-and-burn agriculture

The time-averaged above-ground C stocks for the different categories of land use across the benchmark sites are summarized in Figure 3 and Table 1.  The major conclusions from these data comparisons are:

 

The C stored in the forest systems differed among sites. The highest, >300 t C ha-1, was reported for the undisturbed forests of Indonesia.  It is likely that there were no measurements of primary or undisturbed forests at the other sites.

The C stocks of  the logged forests -- 228, 148, and 93 t C-1hafor Cameroon, Brazil, and Indonesia respectively -- indicate increasing pressure on the forests from farmer-owned logging operations in Cameroon and Brazil to the primarily commercial timber operations in Indonesia. The logged forests contain only about 50% of the above-ground C of the primary forest, while the soil C stocks do not change significantly.

In most places, slash-and-burn clearing is from logged forests, not primary forests.  Using the carbon stored in logged forests (93 to 228 t C-1 ha) as the point from which other systems are derived, the least intensive traditional crop-long fallow systems still practiced in parts of Cameroon have time-averaged C values of 30% of  that of the logged forest.

A slight increase in land-use intensity with the jungle rubber or cacao systems of Indonesia and Cameroon results in maximum C stocks of 90 t C ha-1 and time-averaged C values of  46 to 61 t C ha-1, 30 to 50% of the logged forest.  The time-averaged C stocks of these complex agroforests do not differ much from that of traditional shifting cultivation. The time-averaged C of these systems ranges from 61 t C ha-1, similar to that of the complex agroforests, to 11 t C ha-1 for the coffee plantations in Brazil. The simple agroforests include the multistrata systems of Brazil, that contain two to five major tree species, and the monoculture oil palm, pulp wood, and coffee plantations. The more intensively managed tree plantation systems do not necessarily result in time-averaged C values less than the agroforestry systems.  Industrial plantations may have lower maximum C stocks, but they reach these levels faster than the agroforests, and, therefore, the time-averaged C values can be as high as the agroforests (Figure 2).

5.         The C accumulation rates of the regrowing tree fallow vegetation in Brazil and Cameroon ranged from 3.9 to 8.5 t C ha-1 y-1.  These rates are considerably higher than the 3.0 to 3.6 t C ha-1 y-1 found for the complex cacao and rubber agroforests in Indonesia and Cameroon.  The industrial timber plantations, oil palm plantations, and simple agroforests, however, had relatively high C accumulation rates, ranging from 6.0 to 9.3 t C ha-1 y-1. 

The soil C stocks in the different land-use systems do not change substantially.  Less than 20% of the C in the topsoil (0-20cm) is lost in  agroforestry and tree-based systems.  The largest drops in soil C (50%) are found in some of the degraded pastures and imperata grassland systems.  This 50% drop is equal to a loss of 25 t C ha-1, which is considerably smaller than above-ground losses and potential gains.

The potential for C sequestration in the humid tropics is above ground, not in the soil.  Table 3 provides a summary of the carbon sequestered or lost from the various land-use conversions.  Through the establishment of tree-based systems on degraded pastures, croplands, and grasslands, the time-averaged C stocks in the vegetation increase as much as 50 t C ha-1 in 20 to 25 years, while that in the soil will increase by only 5 to 15 t C ha-1.

Improved fallow systems do not increase time-averaged carbon sequestration substantially from the currently practiced short fallow systems.  This is because of  the short duration time of both the improved and natural fallows, resulting in low time-averaged C stocks.

Improved pasture management does not show an increase in C stocks above ground or in the topsoil compared to the traditional or degraded pastures, at least to levels that would be significant for C sequestration. Fisher et al., (1994) found substantial amounts of C in the roots and subsoil of improved pastures in savannas in Brazil.  Subsoil C was not measured in the ASB plots so there may actually be some storage through improved pastures, although Nepstad et al., (1994) found a dramatic decrease in deep roots on conversion of forest to pasture.

The carbon sequestration potential is overestimated by using the maximum C stored in a land-use system (Figure 2). The correct means of calculating and comparing carbon sequestration is by using time-averaged C stored.

 

 

1. 4 Data limitations, gaps and recommendations for improving the C stock database

The current dataset allows for general comparison of C stocks and time-averaged C values among general land-use types, but some caution must be taken in using these estimates.  There are several steps in which errors may affect the accuracy of the estimates. These include small plot sizes for estimating the biomass of large trees, insufficient numbers of replicates, and inappropriate allometric equations for estimating tree biomass for some of the systems.

 

The total area sampled for tree biomass at each site was 500 m2 (= quadrat size (00m2) multiplied by five (quadrats per site).  Although this may be sufficient in areas where trees are small, < 25 cm diameter at breast height (dbh), it is much less than the 2,500 m2 recommended by Brown et al., (1995) for obtaining accurate measurements in tropical forests where much larger trees are encountered.  The protocol has now been modified to increase the quadrat size to 5m X 100m in areas where there are trees with a dbh  >25cm.

 

The above- and below-ground C estimates for most of the land-use systems were obtained from only three or four true field site replicates (in each field site, estimates were obtained from an average of five quadrats = pseudoreplicates).  In some cases the variability was quite low, but in others it was unacceptably large, and in other cases the estimates were obtained from only two field site replicates. If these C values are to be used for modeling and national inventories, then the accuracy must be improved by increasing the number of replicates.

 

Another source of error could be related to the allometric equation used for estimating the biomass of trees based on their diameter.  The current equation was developed primarily for mature forests that often included only trees greater than 10 or even 25 cm in diameter (Brown et al., 1989).  In addition, the density of the wood in these mature systems may be greater than that in young, regrowing systems.  There are indications that this equation may overestimate the C of trees of dbh < 25cm, which, in fact, includes most of the trees in the secondary forests, fallows, agroforestry and tree plantations measured at the ASB sites. New equations being developed based on extensive sampling of trees in young fallows (Ketterings and van Noordwijk for Indonesia and Palm and Szott for Perú) give estimates half those obtained from the Brown equation.  Several other recent studies have shown a considerable range in allometric equations for both primary and secondary forests in the humid tropics of Brazil (Alves et al., 1997; Araujo et al., 1999; Nelson et al., 1999).

 

Application of these new equations to young, regrowing fallow and agroforestry systems will affect carbon stock estimates and rates of C accumulation. Such systems are currently of interest to the global change community as there is debate on how much C is taken up by regrowing vegetation.  Once new equations for smaller diameter trees and for specific agroforestry species have been agreed upon, then C stocks, C accumulation rates, and time-averaged C values for many of these systems can be improved relatively rapidly.  In addition, since most of the C in these systems is in the trees, we would recommend sampling several more young fallows, mature or growing plantations and agroforestry systems.  The tree biomass will be estimated by measuring dbh of the individual trees, noting which species, and then applying the specific allometric equations.

 

Root sampling and estimation of the C stored in roots has proven to be the most difficult of all the parameters measured.  The estimates for roots have not been included in the tables and figures presented in this report.  If one assumes that the root-to-shoot ratio remains relatively constant for the different systems within a site, then there is a means of estimating the C stored in the root systems.  At the very least, it is possible to say that including roots in the C stock comparisons made above will only magnify the loss of C.  As an example, the roots in a plantation will be less than the roots of a forest system, as is the above-ground C, and therefore the difference in total C between the two systems is larger than, but in proportion to, that estimated by above-ground C only.

 

1. 5 Modeling changes in carbon stocks with changes in land-use

The CENTURY model is a generic plant-soil ecosystem model that can be used to simulate carbon, nitrogen, and phosphorus dynamics of natural and managed ecosystems. Version 4.0 of the model has the ability to simulate complex plant rotations and different types of management practices. The model is well suited for the ASB program because it can simulate the growth of trees and crops and the complex management practices used in the different ASB sites. The various files that have been developed through the collaborative efforts of ASB scientists and the Natural Resource Ecology Laboratory at ColoradoStateUniversity are listed in Table 4. The site files for the different benchmark sites include the basic soil and climate parameters. The other files are then used to construct the sequence of land-use events  (slash-and-burn, crop planting, harvesting, fallow regrowth) that comprise a land-use scenario.

 

Once tested and validated for the benchmark sites, application of the CENTURY model offers opportunity to explore the productivity and carbon losses and sequestration potential of land-use alternatives beyond the time-frame possible from direct field experimentation, for additional land-use systems and for other soil-climate environments. To date, the model has been set up to simulate the carbon, nitrogen, phosphorus and potassium dynamics for the ASB sites in Indonesia, Cameroon, Brazil, and Perú.  The addition of potassium to the CENTURY model was required to simulate crop production on the acid soils of the humid tropics (it could be modified for calcium or other limiting basic nutrients).  This addition of potassium was a conceptual advancement for the CENTURY model.

 

Examples of the simulation modeling and predicting of C stocks for different land-use scenarios can be found in Figures 4 and 5.  The first example shows two of the different land-use systems currently being practiced in Indonesia (Figure 4).  In both cases the forest is first cleared, as indicated by the dramatic drop in biomass C.    In the first case the land is planted to a Paraserianthes pulpwood plantation (see Table 1) with an eight-year rotation.  In the second case, the land is planted to cassava for five or six years, at which time the field is invaded by Imperata cylindrica; after two or three years in cassava the field is recleared and the cycle starts again.

 

The simulation matches the total biomass carbon stocks that have been measured in the field for the tree plantation (left hand y-axis) and the cassava/imperata systems (right hand y-axis).  The biomass carbon simulated for the primary forest is high by about 25%, indicating there may be a need for further model parameterization and validation for the Indonesia site.  The simulated topsoil carbon shows that the tree plantation maintains a steady-state level similar to that of the forest; the blips are a result of the slash that is added and decomposes following tree harvest. Field measurements in the plantation also indicate little or no drop in soil C (Table 2).  The cassava/imperata simulation, however, shows a dramatic and continuing decline in soil C, declining by 40% in 20 years – similar to that from field measurements.

 

The second example from Cameroon simulates the current traditional slash-and-burn agriculture with a declining fallow phase, and two alternative systems.  One of the alternatives includes improvements to the current practice, such as soil conservation and retention of some of the larger trees, and the other alternative makes a switch from the current practice to a tree plantation (Figure 5).  The total system carbon of the forest and long-term fallow are well-simulated, indicating the model has been fairly well parameterized and validated for the Cameroon site.  The alternative system with improved cropping practices shows increases in C stocks compared to that of the traditional system, but the system C still declines, only at a slower rate.  If a tree crop plantation is established (in this case a rubber plantation) the maximum C stocks are similar to the slash-and-burn system, and the time-averaged C stock would actually be higher than the traditional system.  Additional ASB simulations of various land-use systems have been reported for Indonesia in (Sitompul et al., 1997) and Cameroon (Kotto-Same et al., 1997).

The next step for application of the CENTURY Model to ASB will be require careful comparison of model outputs with results of field studies.  A detailed comparison of the model results with the observed data has been run for some of the management practices at the Indonesia and Cameroon sites and needs to be run for all of the sites.

 

1.6 Relating changes in carbon stocks to changes in land-use cover

Data on carbon stocks and time-averaged C stocks of the different land-use systems can be used to determine past, current, and future scenarios of carbon flux with land-use change over larger areas.  Maps of the vegetative land-use cover, based on remote sensing, are available for each of the benchmark areas.  In most cases, vegetation cover maps exist for the areas for at least two points in time.  The changes in land-use cover found in the areas have been related to the changes in carbon stocks associated with the different land-use conversions, as shown in Table 3. Application of this method to three benchmark areas in Sumatra indicated that Rantau Pandan served as a net sink of –3.1 t C ha-1 yr-1 over 64,000 ha for the period 1986 to 1994 (van Noordwijk et al., 1995).  This net sink was due to a large area of regrowing jungle rubber.  The two other areas, however, served as net sources of emissions: 6.8 t C ha-1 yr-1 for Muara Tebo (149,000 ha) and 9.0 t C ha-1 yr –1 for North Lampung (141,000 ha) over the same time period.  An example for the entire state of Rondônia, Brazil shows that the conversion of 93,000 hectares of forest to pasture (a change of 170 t C ha-1) over a 20 year period resulted in a net release of 14 million tons of C to the atmosphere (Fujisaka et al., 1998). A similar example for Cameroon, relates the deforestation of 2 million ha from 1973 to 1988 to a release of 200 million ton of C (Kotto-Same et al., 1997).

These estimates are limited by the errors noted before in the estimates for carbon stocks but also by the difficulty in distinguishing among some of the different land-use covers from remote sensing images. Simply detecting forest clearing for cultivation, when the biggest exchange of carbon with the atmosphere occurs, is fairly straightforward.  But distinguishing logged forests from primary forests and young fallows from jungle rubber may not be possible, while the differences in the carbon stocks and time-averaged carbon stocks among these land-uses can be fairly large.