Williams-Climate-change-refugia-for-terrestrial-biodiversity_0

Table 1: Biological groups compiled for species compositional turnover models.
The number of species and sites are totals prior to site-pair sampling. A different
threshold minimum number of species per site to be included in a model were
identified following preliminary testing. Site-pairs were samples taking into
account the minimum number of species per site.
Taxonomic group
58 Climate change refugia for terrestrial biodiversity
# species # sites # site-pairs
# minimum
species per site
Class Mammalia (mammals) 245 239 162 292 845 ≥5
Class Aves (birds) 599 432 019 296 308 ≥15
Class Reptilia (reptiles) 646 110 221 291 888 ≥3
Class Amphibia (amphibians) 235 75 285 289 683 ≥2
Order Apocrita (wasps and bees) 3840 11 292 297 907 ≥2
Order Coleoptera (beetles) 9288 16 024 297 441 ≥3
Order Aranae (spiders) 2207 10 959 298 138 ≥1
Order Asparagales (lilies and onions) 2308 121 364 296 845 ≥2
Order Asterales (daisies) 2179 208 571 295 982 ≥1
Order Fabales (peas) 3317 299 443 297 145 ≥5
Division Gymnospermae (fruitless seed plants) 212 24 459 294 418 ≥3
Order Myrtales 2564 288 059 294 963 ≥2
Order Poales (grasses and sedges) 2918 266 241 296 777 ≥4
Order Proteales 1159 108 572 294 652 ≥2
Kingdom Fungi 3818 28 156 296 155 ≥2
NSW plant survey 1 4847 40 190 885 376 ≥1
Southeast NSW plant surveys 2 749 13 300 542 361 ≥7
Tingle Mosaic plant surveys 1 773 312 48 205 ≥1
1. Vascular plants – ferns and allies, gymnosperms and angiosperms
2. Trees capable of reaching the canopy (includes gymnosperms and some angiosperms)
4.3.6 GDM model fitting
GDM model fitting followed the procedure outlined in Williams et al. (2012a). Each
correlated group of variables (climate, regolith, landform) was initially tested to identify
redundancy. Remaining candidate predictors were combined (typically around 35–40
variables) and further screened for redundancy using a backward elimination
procedure. Variables were retained if they contributed at least 0.05% partial deviance
explained. Geographic distance between site-pairs was included as a predictor based
on the selection criteria after testing environmental predictors. This procedure resulted
in around 20 predictors in the final model. A detailed description of the GDM model
fitting process and outputs is given in APPENDIX 7. Compositional turnover modelling.
4.3.7 Climate scenarios
Two distinct workflows were applied to the selection and application of climate
scenarios within this study. As an initial test of the refugial analysis, SRES scenarios
were applied to the NSW study region for two GCMS extracted from OzClim, with
standard topographic adjustment. Following development of the complete climatechange
consistent downscaling methodology, two GCMs were selected from the suite
used for the species distribution modelling component of the project, and downscaled