Introduction

The Earth’s surface temperature this century is as warm or warmer than any century since at least 1400 AD (Nicholls et al. 1996). By the year 2100, the average surface temperature is projected to increase by 1.4°–5.8° while sea level is expected to rise by 9–88 cm (IPCC 2001). Greenhouse gases (GHGs) such as (CO2), methane (NH4), nitrous oxides (N2O) and chlorofluorocarbons (CFCs) absorb thermal radiation emitted by the earth’s surface. If more GHGs are emitted into the atmosphere they absorb more heat, which, in turn, could lead to a change in the world’s climate.

Among the GHGs, CO2 is the most abundant and is responsible for more than half the radiative forcing associated with the greenhouse effect (Watson et al. 2000; Schimell et al. 1995). Forest ecosystems play an important role in the climate change problem because they can both be sources and sinks of atmospheric CO2. They can be managed to assimilate CO2 via photosynthesis, and store carbon in biomass and in soil (Watson et al. 2000; Brown 1998; Brown et al. 1996). Great attention is focused on tropical forestry to offset carbon emission due to its cost-effectiveness, high potential rates of carbon uptake, and associated environmental and social benefits (Brown et al. 2000; Moura-Costa 1996; Myers 1996). Tropical forests have the biggest long-term potential to sequester atmospheric carbon (80% of the world’s forests total) by protecting forested lands, reforestation, slowing down deforestation, and agroforestry (Brown et al. 1996). However, at present, the world’s tropical forests are estimated to be a net source of 1.8 Gt of C per year primarily because of deforestation, harvesting and forest degradation (Watson et al. 2000). At least 20% of all atmospheric CO2 emissions are from tropical deforestation.

The Clean Development Mechanism (CDM) of the Kyoto Protocol provides for collaboration between Annex 1 countries and developing countries like the Philippines. The recent meeting of the Parties to the Protocol approved the inclusion of afforestation and reforestation projects under the CDM.

We have earlier estimated the national potential of the Philippines for land-use, land-use change and forestry (LULUCF) projects (Lasco and Pulhin 2001; Lasco and Pulhin 2000). However, there are still no studies on the potential carbon benefits at the project level. The main objective of this study is to analyze the potential of the Magat Watershed for a carbon sequestration project. Specifically, the study aims to: (a) quantify the potential carbon benefits of the watershed, and (b) identify the causes and quantify the potential leakage of the project.

Description of site project

The Magat Watershed is one of the most important watersheds of the Philippines. In recognitions of this, it was declared a forest reservation by virtue of Proclamation Number 573 in June 26, 1969. The watershed supports the Magat multipurpose dam which supplies water for power generation and irrigation as well help control floods.

The watershed is also a dynamic one reflecting the multidimensional problems in most of the country’s upland areas. An open access resource, there is rapid in-migration of people in search of land to cultivate. Vast areas of forests have been denuded resulting to heavy siltation of waterways. The life span of the dam has been significantly shortened by siltation of the reservoir. Floods have also become more frequent.

Biophysical condition

The Magat watershed has a total land area of about 229,000 ha and located between latitude 16o05′ and 17o01′ and longitude 120o51′ and 121o27′. It belongs to the Philippine climatic type III characterized by a lack of pronounced rainfall pattern. It is relatively wet from May to October, with rainfall gradually decreasing from November to February. Total annual rainfall ranges from about 1400 mm in low altitude to about 2400 mm in high altitude.

The dominant soil texture in the watershed is clay loam except along river terraces where silt loam dominates. Upland soils are well drained. Soil fertility is generally high except in grassland areas that have been subjected to high erosion rates. The underlying parent materials are igneous and sedimentary rocks.

The watershed is characterized by a complex web of rivers, streams, creeks and springs. The surface runoff generally follows the rainfall pattern.

Socio-economic Profile

A 1989 census revealed there were 19,375 families in the upland areas. Little is known about the socio-economic profile of the upland dwellers. However, there is an on-going study in the watershed as part of a master planning effort. The results of this study will be available in the next few months.

Methods

The main approach of the study was to quantify the potential net carbon benefits by comparing the baseline scenario with a proposed project scenario. This was done using a combination of primary data gathering in the field and secondary information. In addition, the potential leakage from the project was estimated using historical experience on technology adoption in community forestry projects in the Philippines.

Development of baseline and project scenarios

The study compared two scenarios: a baseline scenario and a project development scenario under a 30-year period (2001–2030). The former is a projection of the condition of the watershed under a business-as-usual scenario while the latter takes into account proposed project activities to rehabilitate and develop the watershed area. Land use and cover data for the last ten years obtained from the National Mapping and Resources Information Authority (NAMRIA) was used to develop the baseline scenario. For the project scenario, the proposed rate of land development was used.

Carbon stocks of study area

The carbon density of the dominant land cover in the watershed was determined using field data collection and secondary data using the methods described by MacDicken (1997) and Hairiah et al., (1998) which we have previously applied in the Philippines. Field data gathering was conducted on August 2001.

For live tree biomass, four 5 × 40 m (200 m2) quadrants were established in each study site except for the Lucban site where only 2 plots were established. Trees >5 cm dbh (1.3 above the ground) that were within 2.5 m of each side of the 40-m centerline were sampled. The species name and dbh were determined, the latter through the use of a diameter tape. If trees >30 cm dbh were present in the sampling area, whether they were inside the 5 × 40 m transect or not, an additional sample quadrant of 20 × 100 m was established where all trees with dbh of >30 cm were measured. The 5 × 40 m quadrant was nested in the lower left hand corner of the bigger plot.

For tree necromass (coarse woody debris, CWD), all dead trees on the ground and dead standing trees >5 cm diameter and >0.5 m in length were sampled by measuring the height/length and diameter at the mid-point.

For understorey biomass and litter layer, destructive sampling techniques were used. Within the 5 × 40 m quadrants, 1 × 1 m and 0.5 × 0.5 m quadrants were randomly laid out for understorey and litter, respectively. Total number of samples for understorey is four while the coarse and fine litter have eight samples each. All vegetation less than 5 cm dbh was harvested within the 1 × 1 m quadrant. The total fresh sample was weighed in the field after which a sub-sample of about 300 g was taken for oven-drying in the University of the College of Forestry and Natural Resources (UPLB-CFNR) laboratory.

Coarse litter is defined as any tree necromass <5 cm in diameter and/or <30 cm length, undecomposed plant materials or crop residues, and all unburned leaves and branches). This was collected in the 0.5 × 0.5 m quadrant on a random location within the understorey sample plot. All undecomposed (green or brown) material was collected and weighed. Similar to understorey, sub-sample of about 300 g was taken for oven drying and carbon content analysis.

After collecting the coarse litter, fine litter is collected in the same 0.5 × 0.5 m quadrant. This was done by collecting about 1000 g of the thoroughly mixed 0–5 cm soil layer (including all woody roots). The sample was then taken in the laboratory where roots and partly decomposed dark litter was dry sieved.

Biomass and necromass samples were oven-dried for 40 hours or until the samples reached their constant oven-dried weight.

About 500 g of soil samples were taken from each of the 5 × 40 m quadrant. The soil samples were obtained at 0–30 cm depth in the 0.5 × 0.5 m quadrant used for litter collection. Bulk density was determined by collecting undisturbed soil cores with a diameter of 5.3 cm and length of 10 cm. The soil samples were initially air-dried and oven dried to constant weight for 40 hours at ±102 °C.

Tree biomass was calculated using the following allometric equation (from Brown 1997):

$$ \eqalign{ & {\text{Y}}({\text{Kg}})\, = \,\exp \,\{ - 2.134\, + \,2.530^{*}\ln ^{*}{\text{D}}\} \,{\text{for}}\,{\text{natural}}\,{\text{forest}}\,{\text{and}}\,{\text{plantation}} \cr & {\text{Y}}({\text{Kg}})\, = \,42.69 - 12.8^{*}{\text{D}} + 1.242^{*}{\text{D}}2\,{\text{for}}\,{\text{natural}}\,{\text{forest}}\,{\text{and}}\,{\text{plantation }} \cr & > 70\,{\text{cm}}\,{\text{dbh}} \cr} $$

The coarse woody debris was also determined using Brown’s equation.

Wood samples were collected from the most dominant tree species in the area. Ground samples were analyzed at the International Rice Research Institute Analytical Service Laboratory (IRRI-ASL) for C content determination using the ROBOPREP C-N Biological Sample Converter. Moreover, available data, based on the previous studies by ENFOR, were used on the carbon content of trees in Makiling and QNP for carbon density determination.

SOC was analyzed using the Walkey-Black method (PCARR, 1980). Total SOC (Mg/ha) = Bulk Density (Mg/m3) *2000 *

%SOC.

Calculation of carbon benefits

Change in carbon stocks over the time was calculated for each land cover type. The net change in each land cover type was aggregated to estimate the total carbon stocks for the whole watershed. The net carbon benefits were determined by subtracting carbon stocks in the baseline scenario from those in the project scenario.

$$ \begin{aligned}{} & {\text{Total}}\,{\text{carbon}}\,{\text{benefits}}\, = \,{\sum {({\text{C}}_{{\text{p}}} - {\text{C}}_{{\text{b}}} )_{{\text{i}}} } } \\ & {\text{where}}\,{\text{C}}_{{\text{p}}} \, = \,{\text{net}}\,{\text{carbon}}\,{\text{stocks}}\,{\text{with}}\,{\text{project}}\,{\text{for}}\,{\text{year}}\,{\text{i}} \\ & {\text{C}}_{{\text{b}}} \, = \,{\text{net}}\,{\text{carbon}}\,{\text{stocks}}\,{\text{at}}\,{\text{baseline}}\,{\text{for}}\,{\text{year}}\,{\text{i}} \\ \end{aligned} $$

where Cp = net carbon stocks with project for year iCb = net carbon stocks at baseline for year i.

Leakage analysis

One of the most critical concerns about forestry projects under the Kyoto Protocol is leakage. The IPCC Special Report on LULUCF defines leakage as the decrease or increase in greenhouse gas benefits outside the project’s accounting boundary as a result of project activities (Watson et al. 2000; Brown et al. 2000). It is more commonly understood in its negative sense, i.e. an anticipated loss of net carbon benefits (Brown et al. 1997). For example, a forest protection project may lead to the cutting of trees in an adjacent forest resulting to minimal net carbon sequestration. Bass et al. (2000) consider leakage as an externality of a project and they differentiated between leakage and slippage. The former occurs when “a project’s activities and outputs create incentives to increase GHG emissions from processes taking place elsewhere”. Slippage occurs when the estimated GHG benefits are negated by an increase in GHG emissions in another area from similar processes. Aukland et al. (2001) divided leakage into two categories: primary and secondary leakage. Primary leakage is synonymous to “slippage” while secondary leakage occurs when the project creates incentives to increase GHG emissions elsewhere.

Recently, an experts’ workshop on LULUCF and the CDM has identified leakage as one of the major research issues (CIFOR 2001). In this study, we analyzed and quantified the potential leakage from the project. The main assumption of the study is that leakage in the project area is largely a function of the degree of adoption of alternative livelihood options of the local communities. In the absence of data from the project site, we conducted a comprehensive literature review of the rate of adoptions of agroforestry and other upland technologies in the Philippines. The values obtained from this review were utilized as a basis for estimating the potential leakage from this project.

Baseline conditions

Baseline land use and cover change

The UMW has a total land area of about 228,000 ha which fluctuates depending on the level water body. Five major land cover types have been identified in the area: built-up/open, (closed-canopy) forest, grassland, non-tree agriculture, and secondary forest/tree plantation (Fig. 1). Built-up areas include human settlements and infrastructure. Forests are presumably closed-canopy forests or old-growth forests. Grassland areas are the result of deforestation and continuous cultivation and grazing. A previous study showed that 20% of all grassland areas are covered with grazing permits. In agricultural area, the favored crops are cereals and sugarcane.

Fig. 1
figure 1

Present land cover of Upper Magat watershed (in ha). Total area 227,919 ha

Full size image

From 1988 to 1998, there had been a marked change in the land cover and landuse of the watershed (Table 1 and Fig. 2). Built-up/open areas registered the greatest relative increase (5.7%) while non-tree agriculture areas (annual crops) had the biggest area decrease (−13%). It is also noteworthy that closed canopy forests increased suggesting recovery of disturbed forests.

Table 1 Historical land cover of Upper Magat watershed (ha)
Full size table
Fig. 2
figure 2

Historical land cover of Upper Magat watershed (1988–1998)

Full size image

On the other hand, secondary forests declined fastest indicating a high deforestation rate in the area. Most of these lands become grasslands while others regenerate back to closed-canopy forests. Similarly, non-tree agriculture is also declining over time. Agricultural lands become grasslands when they are too infertile.

The possible land cover of the watershed for the next 30 years was projected using the average rates of change for the last 10 years (Table 2). This will serve as the baseline scenario for the project. Without intervention, it will be noted that grasslands will continue to increase while secondary forests continue to decline. The combined effect of these two changes will further exacerbate the condition of the watershed. It will mean far greater soil erosion and siltation of water bodies in the watershed.

Table 2 Projected land cover change of Upper Magat Watershed without project intervention (1999–2030)
Full size table

Clearly then, there is a need to arrest the degradation that is happening in the watershed. This project is designed to reverse this downward spiral and promote sustainable watershed development.

Drivers of baseline condition

From the above discussion of land use change, the baseline condition that can be addressed by the project include the loss of carbon in secondary forests and the low carbon density in grassland areas and upland farms (Table 3).

Table 3 Baseline drivers, agents and causes of the baseline condition in UMW
Full size table

Destruction of secondary forests is mainly due to deforestation and small-scale logging activities by small agriculturalists and loggers, respectively. There is no information as to the exact proportion of the annual loss of secondary forests that could be attributed to the two agents.

The low carbon content of grassland areas is first of all due to the lack of economic activity such as tree farming or reforestation. This in turn is brought about by lack of government resources and people who bought the rights to the land because of the hope of using or selling it later. Other causes that perpetuate low carbon stocks are farming and ranching activities.

Similarly, the low carbon stocks of upland farms are maintained by farmers who are dependent of the land for subsistence and cash.

Project development activities

Strategy

The main strategy of the project will be community-based forest management (CBFM). CBFM is a new paradigm that replaced the old mode of forest management in the Philippines. In CBFM, upland communities are given the right to manage and utilize the forest resources on a sustainable basis. It was launched through the issuance of DENR Administrative Order No. 123 on November 28, 1989 to hopefully address the problems of poverty and sustainable management and protection of the remaining second growth forests. Initial experiences on project implementation indicate that given the proper incentives and social and technical preparation, CBFM as a strategy is effective in attaining sustainable forest resource management (La Vina 1997). Successful CBFM sites in the Philippines are listed in Table 4.

Table 4. List of successful community forestry sites in the Philippines (Borlagdan et al. 2001)
Full size table

Assessments made by Borlagdan et al. in 2001 reveal that major impacts of community forestry include: (1) protection of both natural and plantation forests; (2) conversion of grasslands into forests through natural regeneration, tree farms, plantations; (3) biodiversity conservation; and (4) increased productivity of upland farms as a result of gradual restoration of eroded soils.

The main strategy of the project will be community-based forest management. The key actors of the project will be as follows: the local community/PO, an NGO, the DENR, private developer (local firm), project monitoring team, and the investor firm (carbon “buyer”).

All the project activities will be developed with the participation of local communities in the project area. The NGO partner will catalyze the community organizing and development process. The DENR will provide the appropriate land tenure for the project. The private developer will manage and implement the project. The project monitoring team will quantify the carbon sequestered and assess the impacts of the project. The project investor will provide the funds for the project.

Project components and rate of development

The three main components of the project are:

  • •  Protection of secondary forests = 7,430 ha (total area to be conserved in 30 years)

  • •  Reforestation of grassland areas = 100,000 ha

  • •  Agroforestry farms development = 10,000 ha

Protection of secondary forests

Secondary forests are the result of severe disturbance which is typically initiated by logging activities in the Philippines (Lasco et al. 2001). These forests are the most dynamic and economically important forests of the country because they can be exploited through logging. They are also the most vulnerable to deforestation being closer to human settlements and accessible by road.

When these forests are deforested or logged, carbon stored in the biomass is released through oxidation by burning and decomposition. Thus, the main strategy of the project will be to implement protection measures to prevent forest destruction.

The protection of natural forests will be done through a multi-pronged approach. At the local level, farmers will be provided a more stable source of income through agroforestry development so that they will not rely on small-scale logging activities. The organized community will also work with the DENR and local government units in the conservation of natural forests. At the municipal level, a multi-sectoral forest protection committee patterned after the highly successful model in the region will be formed. This committee will be composed of the various sectors of government and civil society. Its primary function is to facilitate coordination and cooperation among the various sectors in the prevention of forest destruction in the target area.

Reforestation of grassland areas

In the Philippines, there are more than 2 M ha of grassland areas. These are not natural but are the results of a degradation process that typically starts with logging of natural forests followed by continuous cultivation and finally grasses like Imperata cylindrica. These grassland areas are maintained by a vicious grass-fire-grass cycle preventing natural succession to proceed. The main approach of the government to rehabilitate these areas is through reforestation activities. However, the success of state reforestation efforts is mediocre at best. This is due in large part to the lack of incentives by major actors to keep the trees alive once the project contract is over (usually after three years).

In contrast, planting trees for carbon sequestration will provide the needed incentives to all actors to keep the trees alive. Since the income stream depends on the existence of carbon stocks, there will be a strong incentive to ensure that they survive.

Reforestation will be done by planting a combination of fast-growing and indigenous species in grassland areas. The fast growing species will be planted ahead to provide shade and improved microclimate for the indigenous species. The species to be planted are: Gmelina arborea, teak (Tectona grandis), pine (Pinus kesiya), narra, and dipterocarp species.

Agroforestry farm development

Upland farms planted to annual crops have high erosion rates and are therefore not sustainable. They will eventually end up as degraded grassland area without any intervention.

Agroforestry will involve the introduction of fruit trees to upland farms that are devoted to annual crops. This will help reduce erosion and increase income of farmers.

As a result of the project activities, the land cover is predicted to change as shown in Table 5.

Table 5 Predicted land area of the UMW under the project scenario with zero leakage
Full size table

Potential carbon benefits

Baseline scenario

To project the land cover change under the baseline scenario, the 10-year average change for each land cover was used (see Fig. 2). Figure 3 shows the predicted land cover under the baseline scenario.

Fig. 3
figure 3

Land cover change in Upper Magat Watershed under the baseline Scenario (1999–2030)

Full size image

Carbon density under each land cover ranges from zero in built up areas to about 200 MgC/ha in closed-canopy forests (Fig. 4). In the next 30 years, carbon stocks will have a net loss of a total of 586,896 MgC (Fig. 5). Most of the carbon will be stored in closed canopy forests.

Fig. 4
figure 4

Carbon stocks of land cover in UMW

Full size image
Fig. 5
figure 5

Net carbon gain/loss from 2001 to 2030 under the baseline scenario

Full size image

Secondary forests carbon will decline the most while grassland carbon will remain steady through time (Fig. 6). On the other hand, closed canopy forest will have a net increase in carbon stored.

Fig. 6
figure 6

Carbon stocks of UMW under the baseline scenario

Full size image

Project scenario

Under the project scenario, the land cover will change dramatically (Fig. 7). The total area of grasslands will progressively decline as a result of massive reforestation efforts. As a result, the area of tree plantations will correspondingly increase.

Fig. 7
figure 7

Land cover change under the project scenario

Full size image

Net carbon benefits will steadily increase to 19.5 Mt C in 2030 (Fig. 8). Of these, the reforestation component will contribute the most increase (Fig. 9).

Fig. 8
figure 8

Net carbon benefits from the project components

Full size image
Fig. 9
figure 9

Total net carbon benefits from project components at 2030, tC

Full size image

Financial costs and benefits

The total project cost for 30 years will amount to US$ 34.5 million (Fig. 10). Most of these will be devoted to the reforestation component. This value is undiscounted (socialized rate is 12% in the Philippines (Lasco and Pulhin 2001).

Fig. 10
figure 10

Total cost of each component (2001–2025), in US$

Full size image

Using the total carbon stocks at Year 30, the cost per ton of C will amount to US$ 1.77. This is within the lower range of the life cycle cost of potential forestry projects in the Philippines (about $ 0.12 per tC to $ 7.60 per tC) (Lasco and Pulhin 2001). It is lower than the cost of protecting a geothermal forest reservation in the island of Leyte (US$ 2.94 per tC) (Lasco et al. 2002).

Estimate of potential leakage

As stated earlier, leakage is the decrease or increase in greenhouse gas benefits outside the project’s accounting boundary as a result of project activities (Watson et al. 2000; Brown et al. 2000). The likely source of leakage for each project component is summarized in Table 6. For the forest conservation component, the main types of leakage will be clearing of adjacent forests by people whose livelihood depended on the preserved forest. Similarly, for the tree plantation and agroforestry farm development components, the main type of leakage is activity shifting by farmers depended on the developed land.

Table 6 Types and causes of likely leakage associated with each of the project component
Full size table

For reforestation, under baseline condition, grasslands are being used for farming and ranching activity. Thus, if these areas are reforested, it would result to the loss of land for farmers/ranchers. This may lead them to clear adjacent forests so that they can continue farming/grazing. For potential agroforestry lands, these are currently utilized for some form of non-sustainable farming. If these are developed for agroforestry, some farmers who don’t adopt the technology may clear forest lands so that they continue their traditional way of farming.

For all project activities, it is noteworthy that the failure to adopt alternative livelihood options is the most important cause of leakage. Thus, the rate of alternative technology adoption is a key factor that will determine the amount of leakage from the project.

In this study, we estimated the potential leakage in the proposed project by taking into account the historical experience in forestry projects in the Philippines. We conducted a literature review of the adoption rate of disseminated technologies and practices in upland watershed areas in the Philippines. Actual rate of adoption ranges from 17–98% with a mean rate of 63% for social forestry projects while agroforestry projects has a mean adoption rate of 71% (Table 7).

Table 7 Rate of adoption of various technologies in upland watersheds in the Philippines
Full size table

It is worthy to note that most of the technologies adopted representing mean adoption rate of 64% are agroforestry, soil conservation measures and social forestry while technology adopted representing low adoption rate of 17% was reforestation. These indicate that local communities are likely to adopt a technology that will augment their income. For instance in the study conducted by Nasayao and Zara (1997) results showed that increased income of the farmers motivated them to adopt agroforestry technology. One advantage of practicing agroforestry is its ability to address the concern of the farmers i. e. ready income from the cash crop or agricultural component of the technology and at the same time its capacity to maintain the ecological stability of the system through its tree component.

Castillo (1991) in her study of farmers adopting agroforestry technology in Ilocos Norte revealed that respondents noted an improvement in their economic status as evident in the changes in the housing materials used (from light materials to semi-concrete).

To increase adoption rate of technology such as reforestation, alternative livelihood should be provided to local communities to reduce their dependence on forest resources. Likewise, studies conducted on adoption noted that factors such as increasing awareness of the local communities of the benefits that can be derived from planting trees (Calanog and Austria 1991; Gerrits et al. 1997, Dolom 1990, Pulhin et al. 2002), security of tenure (Palijon 1988; Chiong-Javier 1997), and size of farms (Chiong-Javier 1997) among others dictate the rate of adoption of the farmers of agroforestry technology. Increased awareness of local communities of the decreased erosion, increased productivity, and sustained water resources due to presence of trees results to higher adoption of the agroforestry technology.

Farmers possessing larger farms are more keen on adopting agroforestry technology because they can devote a certain portion of their land to agricultural crops which can be the source of their immediate cash income while devoting the rest to trees. Local communities with greater security of tenure of the land they are tilling are likely to plant permanent crops such as trees.

Although no study on rate of adoption has been conducted on UMW, rate of adoption from existing studies can be used because upland areas and communities in the Philippines have almost the same characteristics. Upland communities are often characterized by low income, low educational attainment, and dependence on upland farms for their living.

In this study, two leakage scenarios were used: (a) baseline leakage representing the typical rate of adoption for upland development projects; and (b) project case which is based on a high adoption rate (Table 8). The latter is premised on improved adoption as a result of strategies to encourage higher adoption rate such as increased awareness and security of land tenure. These strategies have been shown to be effective in enhancing the rate of adoption in the Philippines.

Table 8 Projected rates of adoption of various technologies in UWM
Full size table

The two leakage scenarios (36% and 83%) have similar and contrasting effects on the carbon stocks of the watershed (Fig. 11). Both of them, resulted to a high leakage in the first ten years of the project. This is because tree planting activities will last up to ten years only. The baseline scenario resulted to a much higher loss of carbon than the moderate leakage scenario.

Fig. 11
figure 11

Amount of carbon lost due to leakage in UMW

Full size image

Amount of leakage was calculated by summing up the annual difference in carbon stocks between the zero leakage and a positive leakage scenario:

$$ {\text{Leakage}}\, = \,\sum {({\text{C}}_0 - {\text{C}}_{\text{L}} )} $$

Total leakage under the project scenario was 3.7 M tC while under the baseline scenario was 8.1 M tC (Table 9).

Table 9 Total leakage (tC) by project component at UMW
Full size table

Summary and conclusions

The study has shown the potential of the UMW to conserve and sequester carbon. The three main development components of the project are forest conservation, tree plantations, and agroforestry farm development. At Year 30, the watershed can attain a net carbon benefit of 19.5 M tC at a cost of US$ 34.5 M.

The potential leakage of the project is estimated using historical experience in technology adoption in watershed areas in the Philippines. Two leakage scenarios where used: baseline and project scenario. Most of the leakage will occur in the first 10 years of the project as displacement of livelihood will occur during this time. The carbon lost via leakage will amount to 3.7 M tC and 8.1 M tC under the baseline and project scenarios, respectively.