英文摘要: | Land-use change, mainly the conversion of tropical forests to agricultural land, is a massive source of carbon emissions and contributes substantially to global warming1, 2, 3. Therefore, mechanisms that aim to reduce carbon emissions from deforestation are widely discussed. A central challenge is the avoidance of international carbon leakage if forest conservation is not implemented globally4. Here, we show that forest conservation schemes, even if implemented globally, could lead to another type of carbon leakage by driving cropland expansion in non-forested areas that are not subject to forest conservation schemes (non-forest leakage). These areas have a smaller, but still considerable potential to store carbon5, 6. We show that a global forest policy could reduce carbon emissions by 77 Gt CO2, but would still allow for decreases in carbon stocks of non-forest land by 96 Gt CO2 until 2100 due to non-forest leakage effects. Furthermore, abandonment of agricultural land and associated carbon uptake through vegetation regrowth is hampered. Effective mitigation measures thus require financing structures and conservation investments that cover the full range of carbon-rich ecosystems. However, our analysis indicates that greater agricultural productivity increases would be needed to compensate for such restrictions on agricultural expansion.
Driven mainly by the fertilizing effects of increased levels of CO2 in the atmosphere, the land system has been a terrestrial sink for carbon in recent decades2. However, the role of land for sequestering carbon is counteracted, as the carbon emissions from land-use and land-cover change accounted for approximately 12% of all anthropogenic carbon emissions from 1990 to 20103. The future development of forest area is uncertain, but deforestation is projected to persist as a significant emission source in the absence of new forest conservation policies, especially under increasing demand for agricultural commodities. Compared to climate change mitigation options in the energy and transport sector, recent research has indicated low opportunity costs and significant near-term mitigation potential through reducing deforestation, promoting avoided deforestation in tropical countries as a cost-effective mitigation option7. Despite the general scientific agreement on environmental benefits of forest conservation, and although the United Nations Framework Convention on Climate Change (UNFCCC) has affirmed the potential role of forests in stabilizing the global climate, no global action has yet emerged to conserve natural forests. Several issues have so far prevented the development of conservation mechanisms supported under the UNFCCC (ref. 8). In particular, the design of financing mechanisms4, but also environmental and socio-political concerns associated with REDD (Reduced Emissions from Deforestation and Degradation) and its variations are being intensively discussed9, 10. One key issue for the implementation of REDD is how to address leakage of emissions11. Without full participation of all countries in a forest conservation scheme, emission reductions in one location could result in increased emissions elsewhere, as agricultural expansion, the main driver for deforestation, could just be displaced rather than avoided12. However, carbon leakage is not only relevant in the context of regionalized forest protection efforts. Another risk associated with a global REDD scheme that so far has not been quantified in the literature is the shift of land-use pressures to non-forest ecosystems (non-forest leakage) simply because they are the only remaining resource for agricultural expansion13. Such ecosystems may also be rich in carbon. First, areas under natural vegetation other than forests, such as shrublands and savannas, can also store considerable amounts of aboveground carbon, especially in Africa, but also in Latin America and Asia6. Second, carbon-rich soils also play a major part in the terrestrial carbon balance and have to be taken into consideration5, 14. Grasslands and pastures, unlike cropland, maintain a permanent vegetation cover and, therefore, have a high root turnover, leading to substantial soil organic carbon storage15. For this reason, carbon stocks decline strongly after land is converted from grasslands and pastures to cropland5. Finally, agricultural activity can reduce carbon sequestration by preventing regrowth of natural vegetation on abandoned agricultural land16. In contrast to the current political discussion, which focuses only on REDD implementation, recent global modelling assessments have focused on the implementation of a global terrestrial carbon policy covering all regions and land types17, 18. To avoid the negative consequences of a global forest conservation policy, a profound understanding of potential implementation failures, such as leakage into land types other than forests, is needed. Here, we estimate land-use and associated carbon dynamics for different global terrestrial carbon policies at global and regional scale using the land-use optimization model MAgPIE (Model of Agricultural Production and its Impacts on the Environment—see Methods)19. Biophysical inputs for MAgPIE, such as agricultural yields, carbon densities and water availability, are derived from a dynamic global vegetation, hydrology and crop growth model, the Lund–Potsdam–Jena model for managed Land (LPJmL; refs 20, 21). LPJmL provides the climate- and CO2-driven changes in carbon densities, agricultural productivity and water availability of a 2 °C scenario (RCP2.6) to drive MAgPIE simulations. For this study, we assume ambitious mitigation policies with different contributions of the land-use sector in three scenarios: no terrestrial carbon policy in the reference scenario (Ref); a global terrestrial land-use policy that considers carbon emissions from deforestation only in the REDD scenario; a global terrestrial carbon policy introduced by a universal carbon tax on greenhouse gas emissions from all terrestrial systems in the All scenario. To account for uncertainty in climate projections, we compute changes in carbon densities, agricultural productivity and water availability for the implementation of the RCP2.6 scenario in five different global circulation models (GCMs). We generally report mean values across all GCMs, while single GCM outputs and standard deviations can be found in Supplementary Table 1. In addition to the default scenarios with different GCM inputs, we perform sensitivity analyses with crucial exogenous parameters (demand for agricultural products, costs for agricultural yield increases and tax on terrestrial carbon emissions) to test the stability of our results in terms of cumulative carbon emissions (see sensitivity analysis in the Supplementary Information). It is important to note that the land-use model not only embraces the calculation of emissions from deforestation and other land-use change, but also the uptake of carbon from regrowth of secondary natural vegetation on abandoned agricultural land and carbon dynamics driven by climate change and CO2 fertilization. In contrast to the mitigation of carbon emissions from land-use change, carbon uptake is not rewarded financially in our scenarios, as we focus in this study on protection policies. The MAgPIE model has been validated intensively for land-use, agricultural yield and land carbon dynamics and reproduces historical trends well (see also the validation section in the Supplementary Information). In addition, the ability of LPJmL to simulate global terrestrial carbon dynamics has been demonstrated in several previous studies21, 22. Our reference scenario (Ref) without any terrestrial carbon policy is parameterized according to the ‘SSP2’ storyline of the shared socio-economic pathways23 (see more detail in Methods). Our model results show that agricultural production increases are mainly realized by intensification on existing agricultural land (Supplementary Fig. 1) as well as by agricultural land expansion. In 2010, global cropland area was 1,454 million ha, pasture land area 3,079 million ha, global forest area 4,144 million ha and global other land area 4,229 million ha (see also Supplementary Fig. 2). At the global level, cropland increases by 237 million ha until the year 2050 and by 239 million ha until 2100, compared to 2010 (Fig. 1). Cropland area expands in developing and emerging regions, including countries of the Middle East and Africa (MAF), countries of Latin America and the Caribbean (LAM) and Asian countries, with the exception of the Middle East, Japan and Former Soviet Union states (ASIA), whereas it decreases in OECD90 countries (OECD; Supplementary Fig. 3). As a consequence, agricultural land is abandoned in the developed regions, as well as in LAM and MAF, where less pasture land is needed owing to more intensified livestock production systems that require less roughage for ruminant feed. Therefore, abandoned land increases by 154 million ha globally until 2100. According to this scenario, global land-use change emissions accumulate to 173 Gt CO2 over the twenty-first century (Fig. 2a). Because of regrowth of secondary natural vegetation, 84 Gt CO2 is sequestered on abandoned agricultural land up to 2100 (Fig. 2b).
| http://www.nature.com/nclimate/journal/v4/n12/full/nclimate2444.html
|