Appendix 4.1

The table below contains detailed information on models and how the upscaling and downscaling were performed for each entry in Table 4.1 (using the same numbering scheme).

In each case E indicates an empirical derivation, M indicates a modelling study, a number refers to how many GCMs (see Glossary) were used in the original literature (for GCM abbreviations used here see below), other codes indicate whether model projections included respectively, precipitation (P), ocean acidification (pH), sea ice (SI), sea-level rise (SLR), sea surface temperature (SST) or anthropogenic water use (W); dispersal assumptions from the literature (D – estimate assumes dispersal; ND – estimate assumes no dispersal; NR – not relevant since species/ecosystem has nowhere to disperse to in order to escape warming – e.g., habitat is at top of isolated mountain or at southern extremity of austral landmass).

IMAGE, BIOME3, BIOME4, LPJ, MAPSS refer to specific models as used in the study, e.g., LPJ denotes the Lund-Potsdam-Jena dynamic global vegetation model (LPJ-DGVM – Sitch et al., 2003; see also Glossary).

Lower case a-h refer to how the literature was addressed in terms of up/downscaling (a – clearly defined global impact for a specific ΔT against a specific baseline, upscaling not necessary; b – clearly defined regional impact at a specific regional ΔT where no GCM used; c – clearly defined regional impact as a result of specific GCM scenarios but study only used the regional ΔT; d – as c but impacts also the result of regional precipitation changes; e – as b but impacts also the result of regional precipitation change; f – regional temperature change is off-scale for upscaling with available GCM patterns to 2100, in which case upscaling is, where possible, approximated by using Figures 10.5 and 10.8 from Meehl et al., 2007; g – studies which estimate the range of possible outcomes in a given location or region considering a multi-model ensemble linked to a global temperature change. In this case upscaling is not carried out since the GCM uncertainty has already been taken into account in the original literature; h – cases where sea surface temperature is the important variable, hence upscaling has been carried out using the maps from Meehl et al. (2007), using Figures 10.5 and 10.8, taking the increases in local annual mean (or where appropriate seasonal, from Figure 10.9) surface air temperature over the sea as equal to the local increases in annual mean or seasonal sea surface temperature. GCM abbreviations used here: H2 – HadCM2, H3 – HadCM3, GF – GFDL, EC – ECHAM4, CS – CSIRO, CG – CG, PCM – NCAR PCM.

The GCM outputs used in this calculation are those used in the Third Assessment Report (IPCC, 2001) and are at 5º resolution: HadCM3 A1FI, A2, B1, B2 where A2 is an ensemble of 3 runs and B2 is an ensemble of 2 runs; ECHAM4 A2 and B2 (not ensemble runs); CSIRO mark 2 A2, B1, B2; NCAR PCM A2 B2; CGCM2 A2 B2 (each an ensemble of 2 runs). Where GCM scenario names only were provided further details were taken from: HadCM2/3 (Mitchell et al., 1995), http://www.ipcc-data.org/ (see also Gyalistras et al., 1994; IPCC-TGCIA, 1999; Gyalistras and Fischlin, 1999; Jones et al., 2005). All used GCMs/AOGCMs have been reviewed here: IPCC (1990), IPCC (1996), Neilson and Drapek (1998), IPCC (2001).