In early studies, the consequences of LUC often were portrayed in terms of the CO2 emissions from tropical deforestation. Early carbon cycle models used prescribed deforestation rates and emission factors to project future emissions. During the past decade, a more comprehensive view has emerged, embracing the diversity of driving forces and regional heterogeneity (Turner et al., 1995). Currently, most driving forces of available LUC-LCC scenarios are derived from population, income, and agricultural productivity assumptions. The first two factors commonly are assumed to be exogenous variables (i.e., scenario assumptions), whereas productivity levels are determined dynamically. This simplification does not yet characterize all diverse local driving forces, but it can be an effective approximation at coarser levels (Turner et al., 1995).
The central role of LUC-LCC in determining climate change and its impacts has not fully been explored in the development of scenarios. Only limited aspects are considered. Most scenarios emphasize arable agriculture and neglect pastoralism, forestry, and other land uses. Only a few IAMs have begun to include more aspects of land use. Most scenarios discriminate between urban and rural population, each characterized by its specific needs and land uses. Demand for agricultural products generally is a function of income and regional preferences. With increasing wealth, there could be a shift from grain-based diets toward more affluent meat-based diets. Such shifts strongly alter land use (Leemans, 1999). Similar functional relations are assumed to determine the demand for nonfood products. Potential productivity is determined by climatic, atmospheric CO2, and soil conditions. Losses resulting from improper management, limited water and nutrient availability, pests and diseases, and pollutants decrease potential productivity (Penning de Vries et al., 1997). Most models assume constant soil conditions. In reality, many land uses lead to land degradation that alters soil conditions, affecting yields and changing land use (Barrow, 1991). Agricultural management, including measures for yield enhancement and protection, defines actual productivity. Unfortunately, management is demonstrably difficult to represent in scenarios.
Most attempts to simulate LUC-LCC patterns combine productivity calculations and demand for land-use products. In this step, large methodological difficulties emerge. To satisfy increased demand, agricultural land uses in some regions intensify (i.e., increase productivity), whereas in others they expand in area. These processes are driven by different local, regional, and global factors. Therefore, subsequent LCC patterns and their spatial and temporal dynamics cannot be determined readily. For example, deforestation is caused by timber extraction in Asia but by conversion to pasture in Latin America. Moreover, land-cover conversions rarely are permanent. Shifting cultivation is a common practice in some regions, but in many other regions agricultural land also has been abandoned in the past (Foster et al., 1998) or is abandoned regularly (Skole and Tucker, 1993). These complex LUC-LCC dynamics make the development of comprehensive scenarios a challenging task.
The outcome of LUC-LCC scenarios is land-cover change. For example, the IMAGE scenarios (Alcamo et al., 1998b) illustrate some of the complexities in land-cover dynamics. Deforestation continues globally until 2050, after which the global forested area increases again in all regions except Africa and Asia. Pastures expand more rapidly than arable land, with large regional differences. One of the important assumptions in these scenarios is that biomass will become an important energy source. This requires additional cultivated land.
Adaptation is considered in many scenarios that are used to estimate future agricultural productivity. Several studies (Rosenberg, 1993; Rosenzweig and Parry, 1994; Brown and Rosenberg, 1999; Mendelsohn and Neumann, 1999) assume changes in crop selection and management and conclude that climate change impacts decrease when available measures are implemented. Reilly et al. (1996) conclude that the agricultural sector is not very vulnerable because of its adaptive capability. However, Risbey et al. (1999) warn that this capability is overestimated because it assumes rapid diffusion of information and technologies.
In contrast, most impact studies on natural ecosystems draw attention to the assumed fact that LCC will increase the vulnerability of natural systems (Peters and Lovejoy, 1992; Huntley et al., 1997). For example, Sala et al. (2000) use scenarios of LUC-LCC, climate, and other factors to assess future threats to biodiversity in different biomes. They explicitly address a biome's adaptive capacity and find that the dominant factors that determine biodiversity decline will be climate change in polar biomes and land use in tropical biomes. The biodiversity of other biomes is affected by a combination of factors, each influencing vulnerability in a different way.
Other reports in this collection