The distribution of the world's vegetation has changed with past changes in climate and will continue to do so in the future. Due to rapidly increasing greenhouse gas concentrations, climate changes now more quickly than it has been doing for a long time  – but the pattern is irregular due to the complex changes in weather patterns, warming and rainfall change. So how much will vegetation change and where will it change most dramatically? Being able to answer these questions, even roughly, is important for two reasons. First, much of human well-being depends on ecosystems, due to the many services they provide . Second, land ecosystems contain large amounts of carbon which could be released as a consequence of major changes – they therefore may accelerate or slow down climate change substantially [3–5] (accelerate, e.g. due to increasing carbon emissions from organic soils, wildfires or forest die-back, or slow down, e.g. through increased vegetation growth and storage in dry or cold soils).
Terrestrial vegetation responds to climate change on several levels. Changes in temperature, precipitation, light and nutrient availability, and in atmospheric CO2 concentration influence plant biochemistry and physiology as well as the allocation of carbon to long- or short-lived plant parts such as leaves, stems and roots. Additionally, plants have evolved different functional strategies to cope with adverse conditions such as drought, cold or inundation (for example, the evergreen and the deciduous strategies of trees), therefore changes in these conditions eventually lead to changes in the species composition of an ecosystem – even if several decades may be needed for the process.
Mapping the outcome of the complex interplay of these processes has become possible due to the development of Dynamic Global Vegetation Models (DGVMs) , which simulate the terrestrial balances of carbon and water as well as the temporal development of vegetation in response to changing climate. The geographical pattern of vegetation emerges as a result of different responses of plant functional types to climate, with respect to productivity, bioclimatic constraints, access to resources and space, and sensitivity to natural disturbances such as fire.
Presently, the land biosphere is a net sink of carbon . Most simulations of the land biosphere's response to future climate change (as simulated by climate models) show a decline in this sink beginning around the middle of the 21st century [4, 8–10], with some scenarios even showing a net carbon loss by the end of the century [3, 9, 11]. The magnitude of this terrestrial feedback on climate is projected to be an additional increase in atmospheric CO2 concentration of between 20 and 200 ppm, implying an additional increase in temperature of between 0.1 K and 1.5 K .
Looking beyond these global numbers, all these simulations contain dramatic regional changes in vegetation structure and composition, in some cases of catastrophic extent. In some model simulations, for example, a collapse of parts of the Amazonian rain forest occurs by the year 2100 due to strongly decreasing rainfall . A decline in boreal forest area due to increasing heat stress on boreal trees has been reported . Another example is a transition from temperate savannah to subtropical woodland for a highland location in Africa . Changes such as these would imply a significant change in the composition and structure of the respective ecosystems – however, they differ depending on the greenhouse gas emission scenario, climate scenario and biosphere simulation model used.
As a step towards the identification of a more robust assessment, we use the state-of-the-art LPJ-DGVM [14, 15] and present results for change from the present in natural vegetation for two scenarios of climate change, selected to represent a wide range of potential futures from moderate (though not weak) to strong change by 2100. The consequences of moderate climate change (temperature increase over land: 2.9 K) were computed for ECHAM5 climate model projections under the SRES-B1 emission scenario (rising to 550 ppm CO2 in 2100). The consequences of strong climate change (temperature increase over land: 5.3 K) were computed for HadCM3 climate model projections under the SRES-A2 emission scenario (rising to 856 ppm CO2 in 2100).