Lambir carbon project

From BioDivBorneo2010

Jump to: navigation, search

New Avenues for Conservation: Carbon Storage in Four Forest Types of Sarawak, Borneo

Cameron Kirk-Giannini, Kimberly O'Donnell and Romadoni Anggoro Paryoto



The creation of a Kyoto-sanctioned carbon credit industry in which companies pay to preserve carbon currently stored in standing forests in order to meet carbon emissions reduction goals promises to both reduce carbon emissions and decrease primary forest habitat degradation. Such a system, however, would necessarily depend on precise measurements of the amount of carbon stored in various forest plots, and such data are presently rare. In this study, we assess the amount of carbon stored in four common forest types (primary dipterocarp forest, heath forest, palm oil plantation and disturbed forest) of the Bornean state of Sarawak. We consider five plots for each forest type, estimating the total above-ground biomass in each plot. The resulting summation of above-ground biomass is then converted into an estimate of the carbon currently stored in each land type. We show that disturbed forests contain significantly less carbon than any of the other three forest types. Primary forest, heath forest, and palm oil plantation contain similar levels of stored carbon on the order of 200-500 tCe/ha. We conclude that disturbance of forest habitat tends to decrease its potential value on the carbon credit market.


Doni in the PDF

As climate change progresses and its effects become more pronounced, proposals for reducing carbon emissions and retaining carbon currently sequestered in natural sources multiply. One influential contemporary effort to control carbon emissions is Europe's CDMA cap-and-trade system, which has effectively created a compulsory market for carbon de novo; there is now a monetary value for sequestered carbon, which translates into a financial value for standing forests. Consequently, it may be possible to use fiscal incentives to promote the conservation and protection of forests. The amount of carbon stored in forests is quite high; previous studies have shown that tropical forests may be storing as much as 40% of the terrestrial carbon and acting as carbon sinks each year they exist[1]. Preserving primary forests is thus a large and potentially lucrative economic strategy for nations with access to these resources. One issue, however, with assessing the carbon balance in any given area is the relatively unknown amount of carbon that fluctuates between the atmosphere and terrestrial environments. This is important information for newly emerging carbon markets because only carbon that is actually being prevented from entering the atmosphere can be applied towards carbon credits. Some attempts to document and record this carbon fluctuation between the land and atmosphere include a report by the US Department of Energy which concluded that from 1850 to 1990, the global net flux of carbon was 124 Pg (1 petagram = 1015 grams) with the greatest flux occurring in South East Asia[2]. The report continues by indicating that the most important factors for such an assessment include meticulous documentation and databases of vegetation, land use, and biospheric carbon content. The report was unable to include how more natural processes such as fire suppression and other environmental factors would affect vegetation. The ability to distinguish this habitual carbon fluctuation from the amount of carbon that is being prevented from escaping due to specific human activity is a pivotal aspect to the integrity of carbon credits in any monetary trading system. Another factor that may influence the carbon credit value of a particular forest type is the amount of carbon being absorbed by the vegetation in the area. Previous data suggests that in times of extensive growth (such as summer or growth after a disturbance), vegetation areas take up more carbon [3]. This could potentially influence the amount of carbon to be stored in different forest types in the future, and thus influence the carbon value of differing land types.

POP example

The state of Sarawak in Malaysian Borneo retains several large tracts of primary rainforest. Although Malaysia's government officially supports conservation (forests are important for “carbon sequestration and oxygen release”[4]), the forests often provide the only source of income for local inhabitants and thus are increasingly being logged and/or converted to palm oil plantations. In fact, it is currently more profitable for local people to convert their land to plantations than to attempt to invest in the voluntary carbon market.[5] If carbon credit prices were on parity with commodity prices in the compliance market, however, their value would increase extensively, possibly making conservation of rain forests more lucrative than deforestation. Thus, if a carbon credit market that valued standing forests were officially established, Borneo and similar nations with primary forests could receive the financial support required to conserve these endangered habitats without damaging local economies. This possibility, however, depends on the availability of accurate measurements of the amount of carbon stored in particular plots of land. Moreover, land use decisions in areas bordering rain forests must be made with relative carbon content measures in mind if the potential financial benefits of carbon storage are to be realized for forest guardians.

DF near Lambir Headquarters

We determine the amount of carbon stored in four different land types in Sarawak. Because of the current issues with logging and increased palm oil plantations, we determined the four land types of interest to be primary dipterocarp forest (PDF), disturbed forest (DF), palm oil plantation (POP), and heath forest (HF). PDF was selected as a control and benchmark against which to compare the others. HF, with its poor, acidic soils and shorter trees, was selected because of its broad importance in Southeast Asian rain forests and its comparative underrepresentation in discussions of conservation and carbon trading. DF and POP were chosen as examples of land uses other than conservation. For the purposes of this study, DF is any land area that was once covered by primary forest but has now been visibly disturbed by logging or other human activities. POP is thus an extreme subset of DF.

Heath Forest at Bukit Pantu in Lambir Hills National Park


  • How do levels of carbon stored in aboveground biomass per hectare differ between primary dipterocarp forests, heath forests, palm oil plantation, and disturbed forests in the Bornean state of Sarawak?


Field methods

  1. A representative site was chosen for each forest type. The PDF site was in the interior of the 52-hectare CTFS Plot in Lambir Hills National Park. The HF was in the area immediately surrounding the summit of Bukit Pantu, also in the park. The POP site was a representative Sarawak oil palm plantation with mature fruiting trees in the vicinity of Lambir Hills. The DF site was a strip of land around Lambir Hills National Park headquarters that had been disturbed by the construction of a nearby highway.
  2. Five 5x10m sample plots were chosen using a randomization procedure in each of the four sites, with two exceptions: the heath forest, due to spatial constraints, could only sustain four plots, and the five PDF plots were 10x10m in size.
  3. When a path was evident, randomization consisted in walking a randomized number of paces down the trail, arbitrarily choosing a direction (right or left), and walking off the trail a set number of paces in that direction. In areas with no path (POP), a transect was randomly defined to take its place.
  4. Diameter at breast height (dbh) and clinometric data (to calculate tree heights) were recorded for all trees in each plot with a diameter greater than 4cm using a tape measure and clinometer (following Dawkins 1961).[6] Clinometric data recorded consisted of: a distance (in meters) from the trunk of each tree to the operator of the clinometer, the angle between the observer's horizontal line of sight and the top of the tree (the total tree height), and the angle between the observers horizontal line of sight and the base of the tree. These three measurements were used to calculate the total height of each tree.


In total, data was collected from over 150 different trees across 19 plots in the four forest types. The quantity of data generated required automated handling for efficiency. Data from each plot was stored as text tables in a field-delimited format in a separate file, and a program was written to import and perform standard computations on the data in each file separately. All data manipulation was performed in the R programming environment. Briefly:

  • Data for PDP plots was corrected for plot size by extracting a random subsample of half of the trees.
  • All angle measurements recorded in degrees were converted to radians.
  • Tree heights were calculated from clinometer data using basic trigonometric methods.
  • Tree volumes were estimated from tree heights and diameters (in meters) according to the formula [1/4 * (d^2) * π] * h.
  • Total plot volumes were calculated by summing the tree volumes of the trees in each plot.
  • Total plot volumes for PDF, SF, and HF plots were adjusted by a factor of 1.1 to account for biomass of foliage and unmeasured plants.[7]
  • A series of t-tests was performed to determine if there were statistically significant differences between the total plot volumes of the different forest types.
  • Tons of carbon dioxide equivalent (tCe) in aboveground biomass per hectare was calculated for plots of each forest type according to the formula (total plot volume) m^3 * 10^6 cm^3/m^3 * 0.6 gdryweight/cm^3 * 0.5 gcarbon/gdryweight * 3.16 gCO2/gcarbon * 10^-6 ton/g * 200 plots/hectare.
  • A record of the R commands used to generate derived data from the stored files is given in the Appendix.


Table 1: p-values of comparisons betwee forest types

The p-values for the t-tests indicate that DF clearly has less above-ground biomass, and consequently fewer tCe/ha, than the three other land types (p < 0.1 in all cases). However, PDF, HF, and POP had statistically equivalent amount of above-ground biomass (0.1 < p < 1). The large amount of above-ground biomass in the POP was especially interesting considering the fact that there was usually only one tree measured per plot. This indicates that the oil palm trees may actually play a large role in carbon storage when at their mature stage, since each tree has the approximate above-ground biomass of all the trees in a similar area in the PDF or the HF.

Figure 1: tCe/ha as a function of forest type
Figure 2: Error bars of data presented in Figure 1

The overall relationship between the four land types was also interesting. For above-ground biomass, the land types display the following relationship: PDF = HF = POP > DF. When viewed with the naked eye, it would appear that the PDF has the most above-ground biomass due to the regular distribution of very tall trees throughout the forest. But the densely-packed small trees of the HF and the single large trees in the POP contained approximately the same amount of wood volume. In fact, as revealed in Figure 1, the POP actually had the most stored carbon of any of the land types studied, with HF storing the second most.


Our results were largely unexpected. We had predicted statistically significant differences between all four forest types, with the relative ranking SF < HF < POP < PDF. Though our results support our a priori placement of the disturbed forest, they do not support any signifcant difference between values for HF, POP, and PDF. The small, thin trees of the HF site had approximately as much above-ground biomass as both the PDF and the POP. In addition, in the HF we noted a profusions of dead matter coating the ground and more young trees below the 4cm cut off mark than the other land types, which suggests that the true above-ground biomass may be even more than recorded despite the correction factor of multiplying by 1.1. Moreover, the single standing trees of the POP represented as much total aboveground biomass as whole plots of other forest types. We tentatively conclude that both heath forest and palm oil plantations are currently undervalued sinks of atmospheric CO2.

There are several possible sources of error in our results. First of all, the small sample size of our plots implies a high variance in our data. This is of special significance in forest types like PDF, where infrequent but massive trees account for the majority of aboveground biomass. Our plots were observed to exclude a number of such large trees, which could have biased our total estimate for PDF downwards. This effect would account for the fact that our calculated tCe/ha for primary dipterocarp forest is slightly lower than comparable values found in the literature. Another possible source of error is the aggregated conversion factors used in the computation of tCe/ha. Wood density was assumed to be invariant between forest types, when in fact it was likely lower in the DF. Similarly, oil palm trees were treated as having uniform wood, when in fact they have an outer coating of low-density wood that contributes to circumference but not tCe. In summary, our estimate for PDF may be artifically low, and our estimates for DF and POP may be artificually high.

In terms of carbon uptake and fluctuation, our data is not sufficient to infer which forest type is currently taking up the most carbon or which forest types are potential sites for clear-cutting. Without this data, we cannot draw conclusions as to which areas are possible revenue sources in terms of carbon trading. The unknown amount of carbon to be taken up in the future by each forest type makes it difficult to asses carbon-credit value. Also, as previously mentioned, carbon stored in forests that are already designated to remain standing is not considered viable carbon for carbon-credit value. A final aspect to be noted is the increased carbon uptake of vegetation in areas with extensive material growth, thus it is likely that forest types which are not in their climax vegetation state are taking up more carbon than those that are. In light of such information, the POP and the SF, which are in states of growth, may have a higher uptake of carbon than the PDF and the HF. The strongest conclusion that can be drawn from our data is the inferiority of DF compared to all other forest types when it comes to tCe/ha. We therefore recommend that policy-makers interested in taking advantage of the burgeoning carbon credit market prioritize ending the clear-cutting of extant forests.


  1. Phillips, O. et al (1998). Changes in the Carbon Balance of Tropical Forests: Evidence from Long-Term Plots. Science, vol. 282: 439-444.
  2. Houghton, R. A., and J. L. Hackler. (2001). Carbon Flux to the Atmosphere from Land-Use Changes: 1850 to 1990. Environmental Sciences Division Office of Biological and Environmental Research U.S. Department of Energy, ORNL/CDIAC-131, NDP-050/R1.
  3. Juday, Glenn, P. et al (2010). Climate change in relation to carbon uptake and carbon storage in the Arctic. The Encyclopedia of Earth
  4. Ministry of Science, Environment and Technology, Malaysia. (April 16 1998). Malaysia's National Policy on Biological Diversity. Government Official Declaration, Kuala Lumpur.
  5. Butler, A. et al (2009). REDD in the red: palm oil could undermine carbon payment schemes. Conservation Letters, Accepted Article.
  6. Whitmore, T.C. Tropical rain forests of the Far East. Oxford: Clarendon Press, 1984.
  7. Ibid.


folnames <- c("pdf", "sf", "hf", "po") ### Import Tables (Format tablex.y)
length(folnames) -> length
for(i in 1:length) {
  for(t in 1:5) {
    assign(paste("table",i,".",t,sep=""),(read.table(paste(t,".csv", sep = ""), 
      sep = "|", header = T)))
num <- c(1,2,3,4) ### Calculate Tree Volumes
for(i in 1:length(num)) {
  assign("zmatch",grep(paste(i,".",sep=""), objects()))
  for(t in 1:5) {
    (objects()[zmatch[t]] -> working)
    assign(working, subset(get(working), diam >= 4))
    get(working)$alpha*3.1415926/180 -> tempalpha
    get(working)$beta*3.1415926/180 -> tempbeta
    tan(tempalpha)*get(working)$dist -> temph1
    tan(tempbeta)*get(working)$dist*-1 -> temph2
    temph3 <- temph2 + temph1
      * .0001*get(working)$diam*.25*3.1415926)*0.5) 
    ### Store Tree Volumes (Format voltablex.y)

for(i in 1:length(num)) { ### Calculate Plot Volumes
  assign("zmatch",grep(paste("voltable",i,".",sep=""), objects()))
  for(t in 1:5) {
    (objects()[zmatch[t]] -> working)
    assign(paste("xtot",i,".",t, sep=""),sum(get(working))) ### Store Plot Volumes (Format xtotx.y)
xtot1 <- c(xtot1.1,xtot1.2,xtot1.3,xtot1.4,xtot1.5) ### Concatenate Plot Volumes (Format xtotx)
xtot2 <- c(xtot2.1,xtot2.2,xtot2.3,xtot2.4,xtot2.5)
xtot3 <- c(xtot3.1,xtot3.2,xtot3.3,xtot3.4,NA)
xtot4 <- c(xtot4.1,xtot4.2,xtot4.3,xtot4.4,xtot4.5)
xtot1 <- 1.1*xtot1 ### Adjust for Excluded Wood
xtot2 <- 1.1*xtot2
xtot3 <- 1.1*xtot3
t.test(xtot1,xtot2)  ### Perform T-tests
xtot1*.6*.5*3.1415926*200 -> tce1 ### Calculate tCe/ha
xtot2*.6*.5*3.1415926*200 -> tce2
xtot3*.6*.5*3.1415926*200 -> tce3
xtot4*.6*.5*3.1415926*200 -> tce4