Integrated assessment of options to reduce emissions from deforestation in Tshopo, DR Congo
Mitigation climate change is an urgent and crucial challenge for a sustainable earth-system. Curbing tropical deforestation, and increasing absorption of carbon by (new) tropical forests is being considered as a relevant mitigation strategy because of the assumption that it can be implemented quickly and more cost-effective than other strategies. Reducing Emissions from Deforestation and Forest Degradation (REDD+) has become one of the initiatives most prominently present in research and international climate policy. The idea is to create incentives for developing countries and forest users to reduce carbon emissions by keeping their forests standing, reducing forest degradation, managing forests sustainably and increasing forest carbon stocks. However, in order to make the mechanism work and to design effective, efficient and equitable REDD+ policies, adequate information is needed on several aspects of the local land system where one wishes to intervene: this entails the current land use and land use evolutions, its drivers, and the socio-economic and carbon outcomes of different land use systems. Despite its importance in terms of forest cover and rising deforestation rates, the Democratic Republic of Congo has been underrepresented in studies that address these aspects, especially at subnational scales. In this thesis, we aimed at achieving a better understanding of two aspects of the shifting cultivation land system in Tshopo province: (1) the effects of biophysical conditions, land use and land management on ecosystem services delivery, in particular carbon sequestration, and (2) the drivers of the evolutions of land use and land management, including options for future transitions.
In chapter 3, we analysed variation in aboveground carbon and forest structure in old-growth forests in the central Congo Basin, and assessed its implications for carbon mapping. We found that aboveground carbon (AGC) differed significantly within the region and between forest types. With an estimated aboveground carbon stock of 189 ± 40 Mg C ha-1 mixed forest stored significantly less carbon than monodominant forest, storing 225 ± 58 Mg C ha-1. The difference in AGC between the two forest types was due to differences in mean wood density and mean tree heights and not basal area. Our observations confirm generally lower aboveground carbon stocks in the central Congo Basin as compared to the outer fringes. Environmental drivers of AGC correlate to a large extent to environmental conditions determining the occurrence of the two forest types, with higher AGC on sandier soils, but drivers of AGC in mixed forest remain inconclusive. Taking differences in forest structure across forest types into account will benefit forest carbon mapping at different scales. Full inventory plot AGC estimations can benefit from improved accuracy when using forest type specific height:diameter relationships. We further tested the capacity of Asner’s LiDAR universal model in predicting regional level AGC in old-growth forest and found that AGC could be predicted with an 80 % precision. We also found that when differences in forest type are not taken into account, a forest type bias would be induced.
In chapter 4, we evaluated the importance of management history and plant community properties on biomass stocks and biomass productivity in fallow systems in the central Congo Basin. The sampled fallows (aged 8.7 ± 5.5 years) had 62.5 (± 47) Mg ha-1 aboveground dry biomass (AGB), on an average. AGB was positively related to fallow age and negatively related to the number of previous cultivation cycles. Fallows in the fourth cycle stored significantly less carbon. We explored the relations between management history, landscape characteristics, woody plant community properties and annual biomass increment (ABI) in fallows aged 5 to 10 years. Three hypotheses on how woody plant community properties could affect ABI were tested: functional diversity, functional identity and vegetation quantity. Stem density was the strongest predictor of mean annual biomass increment (ABI), while also the community weighted mean of traits associated with pioneer species (low WD and low adult stature) was positively related to biomass recovery rates. Management history strongly determined these community properties, with the number of previous cultivation cycles negatively affecting stem density and community-weighted pioneer trait values, and with the biomass of remnant trees negatively related to community-weighted pioneer trait values. Our results imply that management intensity, i.e. mainly the number of previous cultivation cycles associated with repeated burning and weeding and with Chromolaena odorata invasion leads to a decrease in biomass recovery, with the main effect occurring through a reduction in the density of regenerating individuals.
In chapter 5 we assessed the variation in deforestation rate and drivers at different scale levels, among which the village and household level. We detected considerable variation in deforestation rate at each scale levels, with highest variation at village and household level, which has gone unaddressed so far in DRC’s REDD+ strategy and projects. Village level deforestation rates were best predicted by the distance to Kisangani, the distance to the forest, the village ethnicity and the village population density. Individual household contributions to deforestation were also strongly unevenly distributed. Our results demonstrated that not the poorest households, but those with certain means in terms of social and human capital are more likely to participate in primary forest clearing. About half of the deforestation was attributed to only 8% of households. For those households, deforestation is part of their development strategy, with both short and mid-term benefits (increased harvests), and long-term benefits (ownership over lands to be passed on to their children).
In chapter 6 we examined to what extent the concept of local forest cover transitions can provide a framework to identify appropriate land use interventions. Using land cover, population and population density and the importance of different livelihoods, we identified three village level forest transition types: (i) high forest cover-low population, (ii) low forest cover-low population and (iii) low forest cover-high population. Perceptions of forest cover change and the evaluation of forest availability vary between these types. In general, forests were mostly valued for provisioning services and as a land reserve for future agricultural development. The most valued forest products were perceived to be in decline and scarce (except wood products), irrespective of the local forest transition type. We conclude that (i) local ownership of the purpose of forest conservation is not met in a majority of villages due to a perception of abundant forest cover, (ii) restricting forest conversion under REDD+ without readily available, acceptable and better functioning alternatives could result in maladaptive changes in land and resources management and negative livelihood and environmental outcomes, and (iii) interventions supporting higher agricultural outputs per ha will be best received in low forest-high population villages.