the Arctic and montane forests). The current implementa­tion in the model is based on changing temperature only. Estimates from a European model of the proportion of spe­cies lost per biome (Bakkenes et al., 2002; Leemans and Eickhout, 2004; Bakkenes et al., 2006) for increasing levels of temperature are applied within the GLOBIO3 model on a global scale. This regional bias and the absence of a mod­eled response to changes in moisture availability are impor­tant areas for model improvement.
     Some responses to change take some time to become apparent. The loss of species from a particular area may take 30 years or may be instantaneous, depending on the type and strength of the pressure. Because of these lags, the model outcome portrays the possible impact over the short to medium term (~5 to 30 years). These lags must be bet­ter characterized, and for that the underlying databases are developed further.
     There is little quantitative information about the inter­action between pressures. Various assumptions can therefore be included in the model, ranging from "all interact" (only the maximum response is delivered) to "no interaction" (re­sponses to each pressure are cumulative). The GLOBIO 3 model calculates the overall MSA value by multiplying the MSA values for each driver for each IMAGE 0.5 by 0.5 degree grid cell according to:

MSAi = MSALU MSAN MSAI MSAF MSACC

where i is the index for the grid-cell, MSAXi relative mean species abundance corresponding to the drivers LU (land cover/use), N (atmospheric N deposition), I (infrastructural development), F (fragmentation) and CC (climate change). MSALUi is the area-weighted mean over all land-use cat­egories within a grid cell.

     The model relates 0.5° IMAGE maps to Global Land Cover 2000 as a base map at a 1-km scale, based on a series of simple decision rules. These maps are used to estimate the response to changes in land cover and land use inten­sity within each 0.5° grid cell. The land-use cover maps and the maps representing other pressures are used to generate maps of the share of remaining biodiversity, which may be derived either in terms of remaining share of original spe­cies richness, or remaining share of mean original species abundance. More data is being collated for abundance than for richness—this is the favored indicator, as it is closest to those specified by CBD. Outputs are derived at a 0.5° scale and can be scaled up to IAASTD regions.

A.5.8.3 Application
GLOBIO3 has been used in global and regional assess­ments. GLOBIO3 analyses contributed to an integrated as­sessment for the Himalaya region (Nellemann, 2004); for deserts of the world and the Global Biodiversity Outlook (SCBD/MNP, 2007).

A.5.8.4 Uncertainty
GLOBIO3 reflects a relatively new model approach. The level of confidence is highly related to the data quality and quantity, a lot of which is derived from other models, in par­ticular, the IMAGE 2.4 model, infrastructure maps (Digital Chart of the World or DCW) and other land cover maps. The biodiversity indicator generated (MSA) is designed to be compatible with the trends in abundance of species in­dicator as specified by CBD. Other indicators might lead to different results. However the patterns of the global analyses are in line with earlier global analyses. Table A.5.14 provides an overview of major parameters and model structure.