Williams et al. 10.1073/pnas.0606292104. |
Fig. 5. Maps of novel and disappearing climates generated by using a global search and color scaling based on a critical SED threshold of 5.33. The format follows Fig. 2.
Fig. 6. Maps of novel and disappearing climates generated by using a 500-km search restriction and a color scaling based on a critical SED threshold of 5.33. The format follows Fig. 3.
Fig. 7. (A and B) Box plots showing the fractional global area with local climate changes that exceed an SED threshold (A: SEDt = 3.22; B: SEDt = 5.33) equivalent in scale to the differences in 20th-century climatic regimes among potential vegetation types. Local climate changes that exceed SEDt may be large enough to induce biome-scale vegetation change. Box plots show the range of projections among the nine A2 and eight B1 experiments analyzed here. The median projection is indicated by the horizontal line intersecting each box. Upper and lower box limits indicate the 25th and 75th percentiles among model projections and whiskers indicate the uppermost and lowermost projections. (C and D) As A and B, but applied to the SED's between the climate vector for each 21st-century gridpoint and its closest 20th-century regional analog, for the A2 and B1 scenarios and for analog searches conducted globally and within 500 km of each gridpoint. Here SEDs >SEDt indicate biome-scale climate changes with no 20th-century analogs within the search domain. (E and F) As C and D, but applied to the SED's between the climate vector for each 20th-century grid and its closest 21st-century analog. Here SEDs >SEDt indicate the disappearance of the 20th-century climate regime from the search domain.
Table 1. Models
Model Name | Modeling Group | Refs. |
CCSM3 | National Center for Atmospheric Research | 1 |
CSIRO-Mk3.0* | CSIRO Atmospheric Research | 2 |
ECHAM5/MPI-OM | Max Planck Institute for Meteorology | 3 |
GFDL-2.1 | U.S. Department of Commerce/NOAA/Geophysical Fluid Dynamics Laboratory | 4 |
GISS-ER | NASA/Goddard Institute for Space Studies | 5 |
IPSL-CM4 | Institut Pierre Simon Laplace | 6 |
MRI-CGCM2.3.2 | Meteorological Research Institute | 7 |
PCM | National Center for Atmospheric Research | 8 |
UKMO-Had3 | Hadley Center for Climate Prediction and Research/Met Office | 9 |
*Only the SRES A2 scenario was used from CSIRO-Mk3.0; the SRES B1 was not available at the time of analysis.
1. Collins WD, Bitz CM, Blackmon ML, Boman GB, Bretherton CS, Carton JA, Chang P, Doney SC, Hack JJ, Henderson TB, et al. (2006) J Clim 19:2122-2143.
2. Delworth TL, Broccoli AJ Rosati A, Stouffer RJ, Balaji V, Beesley JA, Cooke WF, Dixon KW, Dunne J, Dunne KA, et al. (2006) J Clim 19:643-674.
3. Gordon C, Cooper C, Senior CA, Banks HT, Gregory JM, Johns TC, Mitchell JFB, Wood RA (2000) Clim Dynamics 16:147-168.
4. Gordon HB, Rotstayn LD, McGregor JL, Dix MR, Kowalczyk EA, O'Farrell SP, Waterman LJ, Hirst AC, Wilson SG, Collier MA, et al. (2002). The CSIRO Mk3 Climate System Model (CSIRO Atmospheric Res, Aspendale, VIC, Australia), Report 60.
5. Marti O, Braconnot P, Bellier J, Benshila R, Bony S, Brockmann P, Cadule P, Caubel A, Denvil S, Dufresne JL, et al. (2005) The New IPSL Climate System Model: IPSL-CM4 (Inst Pierre Simon Laplace, Paris), Report 26.
6. Roeckner E, Bäuml G, Bonaventura L, Brokopf R, Esch M, Giorgetta M, Hagemann S, Kirchner I, Kornblueh L, Manzini E, et al. (2003) The Atmospheric General Circulation Model ECHAM5 (Max Planck Inst for Meteorol, Hamburg, Germany), Report 349, Part I.
7. Schmidt GA, Ruedy R, Hansen JE, Aleinov I, Bell N, Bauer M, Bauer S, Cairns B, Canuto V, Cheng Y, et al. (2006) J Clim 19:153-192.
8. Washington WM, Weatherly JW, Meehl GA, Semtner AJ, Jr, Bettge TW, Craig AP, Strand WP, Jr, Arblaster JM, Wayland VB, James R, Zhang Y (2000) Clim Dynamics 16:755-774.
9. Yukimoto S, Noda A, Kitoh A, Sugi M, Kitamura Y, Hosaka M, Shibata K, Maeda S, Uchiyama T (2001) Papers Meteorol Geophys 51:47-88.
Table 2. SEDt for individual biomes for a T42 model grid, as determined by ROC analysis
Biome | SEDt | Area, 1,000 km2 |
Tropical and Subtropical Moist Broadleaf Forests | 5.216 | 1,653 |
Tropical and Subtropical Dry Broadleaf Forests | 7.488 | 312 |
Tropical and Subtropical Coniferous Forests | 11.103 | 62 |
Temperate Broadleaf and Mixed Forests | 3.564 | 1,467 |
Temperate Conifer Forests | 4.253 | 474 |
Boreal Forests/Taiga | 1.931 | 2,483 |
Tropical and Subtropical Grasslands, Savannas, and Shrublands | 3.793 | 1,647 |
Temperate Grasslands, Savannas, & Shrublands | 2.553 | 1,160 |
Flooded Grasslands & Savannas | 4.394 | 106 |
Montane Grasslands & Shrublands | 5.236 | 494 |
Tundra | 1.978 | 2,959 |
Mediterranean Forests, Woodlands & Scrub | 6.158 | 322 |
Deserts and Xeric Shrublands | 2.480 | 2,625 |
Mangroves | 17.514 | 29 |
AREA-WEIGHTED AVERAGE | 3.217 | |
AREA-WEIGHTED STANDARD DEVIATION | 2.115 | |
AVERAGE + 1SD | 5.332 | |
AVERAGE - 1SD | 1.101 |
SI Materials and Methods
SEDt varies by biome, because biomes differ in the range of climatic conditions they occupy. SEDt tends to be larger for minor and dispersed biomes (e.g., Mangroves, Tropical/Subtropical Conifer Forests; SI Table 2) because they tend to have less cohesive climatic distributions and therefore higher SEDt. SEDt for individual biomes ranged from 1.93 to 17.5 (SI Table 2), with an area-weighted average and standard deviation of 3.22 ± 2.11 (the range of SEDt is inflated by high values for minor biomes such as Mangroves and Tropical and Subtropical Conifer Forests; see SI Table 2). Because the interbiome variation in SEDt is partially an artifact of the size and extent of individual biomes, we here use the area-weighted global average SEDt instead of applying biome-specific SEDt values to determine whether a particular SED score represents a novel or disappearing climate.
Altering the value of the global SEDt does not change the mapped patterns shown in Figs. 2 and 3, but it does affect the areal percentages of novel and disappearing climates (Fig. 4). To assess the sensitivity of our results to choice of SEDt, we reanalyzed our data using a more conservative value of 5.33 (i.e., the mean plus one standard deviation). As expected, the predicted extent of novel and disappearing climates decreases when a more conservative threshold is applied (SI Figs. 5-7). The fundamental mapped patterns (SI Figs. 5-7) remain unchanged, but the extent of areas classified as having novel or disappearing climates (SI Fig. 7) decreases. The projected extent of novel climates is 1-17% (A2) and 0-4% (B1) with no dispersal limitation and 16-61% (A2) and 1-20% (B1) with a 500-km dispersal limitation (SI Fig. 7). The projected extent of disappearing climates is 0-16% (A2) and 0-2% (B1) with no dispersal limitation and 15-61% (A2) and 1-18% (B1) with a 500-km dispersal limitation (SI Fig. 7). Even with this conservative threshold, 21st-century climates with no 20th-century analog anywhere globally are projected to occur in the Amazonian and Indonesian rainforests, and disappearing climates are projected for the central Andes and eastern Africa (SI Fig. 5). The projected occurrence of novel and disappearing climates remains pervasive when dispersal limitations are also accounted for (SI Figs. 6 and 7).
One risk of using a global SEDt, however, is that the interbiome difference in SEDt may reflect a real difference among biomes in the sensitivity of their constituent species to climate change. If so, then our use of a global threshold (SEDt = 3.22) would tend to overestimate the ecological impacts of climate change for biomes with a high SEDt (e.g., Tropical and Subtropical Moist Broadleaf Forests) and underestimate the ecological impacts of climate change for biomes with a low SEDt (e.g., Boreal Forests/Taiga) (SI Table 2). Because the global SEDt used here is an area-weighted global average, using a biome-specific SEDt would be unlikely to drastically change the areas of novel and disappearing climates calculated here (SI Fig. 7), although it would alter the assessment of which regions are at greatest risk.