Assessment of watershed functions to support negotiations in a catchment under land use conflict in Sumberjaya, Lampung, Sumatra, Indonesia

Deforestation is often blamed for having tremendous negative impacts on watershed functions. The watershed functions delivered by forests are generally understood to be reduced peak flows, greater dry season discharge volumes, landslide and water erosion prevention, improved water quality and reduced sedimentation of reservoirs and waterways. These functions are commonly used as the main justification for classifying large areas of land as protection forest. In Indonesia alone, 15 % of the total land area has been classified as ‘protection forest’, which is 20 % of the 143 million ha designated as State Forest Land. Nevertheless, much of the forest cover has already disappeared in State Forest Land. In Indonesia, the classification of land as protection forest is based on criteria such as slope, altitude, rainfall and soil, but not on criteria directly linked to the above-mentioned watershed functions. Communities that live in the upper areas of watersheds are frequently blamed by government agencies for causing a loss of watershed functions and are asked to alter their land-use practices or to give up their land altogether. In the worst cases, farmers have been evicted when large-scale reforestation projects have been implemented, e.g. when dams are constructed. Sumberjaya, a mountainous area in West-Lampung, Sumatra, Indonesia was chosen as a representative site for the above described situation. Forest cover in the upper Way Besai catchments decreased from 46 % in 1974 to slightly more than 12 % in 2002. Most forest was converted to coffee gardens, which were blamed as the cause of catchment degradation. Violent evictions with military support took place between 1991 and 1996. This heavy-handed command-and-control approach did not restore any forest. After the fall of the Suharto regime in 1998 and the implementation of local autonomy in 1999 a different approach had to be tried out, as the Way Besai hydropower dam was already under construction. A negotiation process was started between the most important stakeholders, among them foresters, farmers and managers of the hydropower dam. Various research activities were carried out by ICRAF and partners to characterise the area, to try to answer specific research questions and to support the ongoing negotiations. Aim of this thesis is to assess how changing agroforestry landscape mosaics in Indonesia affect watershed functions. Key hypothesis in this current research is that some farmer-developed agroforestry mosaics are as effective as the original forest cover to sustain desired watershed functions, and hence that conflicts between state forest managers and local population can be resolved to mutual benefit. We first sketch how different perceptions regarding watershed functions exist between various stakeholders with their respective knowledge domains. The schematical ‘environmental issue cycle’ offers an apt theoretical framework to show how public perceptions evolve over time through social interaction and scientific enquiry. A state-of-the art description of the most dominant processes affecting watershed functions is presented (Chapter 1). This is made more concrete with a description of the study site from both a biophysical, as well as a socio-economic perspective (Chapter 2). A study of past land use changes and their causes shows that the past forest conversion was the result of many driving factors, situated mainly outside the forestry (policy) domain. All proximate causes (infrastructure extension, agricultural expansion and wood extraction) and underlying driving factors (population growth, economic factors, technological improvements, the political and institutional setting and cultural environment) reinforce each other and create a very strong deforestation-oriented context. After a peak of deforestation in the mid-seventies, a phase of regreening or ‘retreeing’ started since the mid-eighties, while deforestation was still going on. More and more shade trees were introduced in the coffee gardens and various shade coffee systems developed and evolved on the undisputed private land, where secure land tenure was not an issue, but this did not happen in the State Forest Land (Chapter 3). A compilation of available soil data combined with the available land use data shows that the highest infiltration capacity, highest soil organic carbon content and the lowest soil bulk density were all measured under forest. After a phase of degradation soil physical properties can however improve over time in coffee agroforestry systems as more trees are being introduced. More research is needed on why the recovery of soil physical properties in Sumberjaya is seemingly slower than would be expected from available literature. This can be due to the on average still low number of shade trees, the choice of tree species and irreversible changes in soil texture. More research is needed to e.g. assess to what extent the choice of tree species has an impact on the litter layer and the soil fauna, which again affects the soil physical properties (Chapter 4). Over the past 30 years average yearly rainfall shows a slight, but insignificant decreasing trend in the Sumberjaya area. The seasonality did however increase significantly (albeit at a low level, p=0.87) and shows a high correlation with the Indian Ocean Dipole and the Oceanic Niño Index. This suggests that especially in dry years, sea surface temperatures have a much larger impact on the rainfall patterns than local land use changes. Studies in other tropical areas however show that besides sea temperatures, also other factors like the size of the islands play an important role, because of their impact on the magnitude of the sea breeze. Forest does have an important influence on rainfall patterns in the maritime tropics, like Indonesia. More research is needed in order to quantify the importance of the various factors (Chapter 5). Parsimonious rainfall-runoff modelling is a powerful tool to combine independent rainfall and discharge datasets. A thorough data quality analysis is needed to identify and separate good quality data from very poor data. Over the past 30 years the runoff coefficient increased on average with ca. 7 %. Most of this increase (6 %) is attributed to reductions in interception and evapotranspiration losses after forest conversion to a mosaic of coffee systems on the hill slopes with paddy rice in the valleys. A part (1 %) is attributed to the global increase of CO2, affecting the transpiration efficiency of plants worldwide. Claims that the average annual runoff discharge reduced over time were largely based on the poor quality of data collected over the last 10 years. Variability in annual discharge is in the order of 1200 mm, which is due to the variability in annual rainfall. This is 4 times more than the range that can possibly be attributed to land use change (300 mm). The decreasing trend in dry season flows must largely be attributed to an increased seasonality in rainfall and the occurrence of longer consecutive dry periods and not to the past changes in land use. The year-to-year variability in dry season flow during the two driest months of the year that is due to the variability in rainfall, is in the order of 190 mm. The modelling illustrated that the impact of the past land use changes (like a reduced infiltration capacity) only has an impact in the order of 8-25 mm (Chapter 6). Sediment yield assessments at the catchment level preferably require a continuous series of hydrological data and discrete sediment concentration data. Gaps in runoff discharge data can be filled using parsimonious rainfall-runoff models like IHACRES or VHM. Peak flows are generally responsible for the bulk of the sediment transport. It is thus of the utmost importance to correctly predict these peak flows in order to accurately estimate sediment yield. The VHM model showed a higher accuracy in predicting peak flows in function of rainfall than the IHACRES model. It also illustrated how the floodplain in the southern half of the upper Way Besai catchment effectively reduced these peak flows and the associated erosive power (Chapter 7). Sediment yield is very variable between the various subcatchments in Sumberjaya. A multivariate analysis illustrates that the most important explaining factors are differences in the underlying geology, slope and the silt fraction of the soil. Catchments on the northern slope of the Bukit Rigis represent less than 20 % of the area, but contribute almost 60 % of the suspended sediment. Plot level erosion measurements show that on lithological sensitive areas, erosion peaks in the third to fourth year after deforestation and can be as high as 16 Mg ha-1 yr-1. With time erosion reduces below a level of 5 Mg ha-1 yr-1. Sediment yield was 3 to 10 times higher at the catchment level compared to data collected at the plot level. This suggests that surface erosion is not the most dominant process in Sumberjaya. Landslides, river bank and river bed erosion and the concentrated flow erosion on footpaths and dirt roads are the most important erosive processes. Sediment concentration data taken at the same location covering two thirds of the catchment in 1989-1990 and in 2005-2007 yielded practically the same sediment rating curve, suggesting that there is no difference in sediment yield over time. On the other hand the 2005-2007 samples taken further downstream showed an increase in sediment concentration, compared with 1989-1990 data. This difference could be due to changes in the catchments on the northern slope of Bukit Rigis, catchments that are situated in between the two measuring points, but also partly to the exclusion of the most sediment rich samples in the 1990 sediment rating curve (Chapter 8). Coffee agroforestry systems are gaining increased acceptance in forestry circles as a sustainable and profitable land use system, despite an institutional context that is still embedded in a forestry – agriculture dichotomy. In 2001 the first experimental social forestry (HKm) agreements between local farmer groups and the Forestry Department were signed leading to the beginning of a solution of the past land use conflicts. Farmers obtained the right to grow coffee in state forest land on the condition of protecting the remaining forest and planting shade trees in the coffee gardens. More trees are now appearing in the monoculture coffee gardens and the sometimes bare land in State Forest Land. The remaining forest is now effectively being protected, as since 2007, for all the protection forest in Sumberjaya, there are now signed HKm agreements with farmer communities. The HKm agreements will lead to a further increase in tree numbers and soil conservation measures (e.g. terracing, reduced weeding). This will lead to a reduced surface erosion, but will likely only have a marginal impact on the sediment yield at the catchment level as the most important erosive processes (landsliding, bank erosion and concentrated flow erosion on footpaths and dirt roads) are not included in the agreements. Alternative approaches to reduce the sediment yield at catchment level, need to be explored (Chapter 9).