Premium
The ice age ecologist: testing methods for reserve prioritization during the last global warming
Author(s) -
Williams John W.,
Kharouba Heather M.,
Veloz Sam,
Vellend Mark,
McLachlan Jason,
Liu Zhengyu,
OttoBliesner Bette,
He Feng
Publication year - 2013
Publication title -
global ecology and biogeography
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 3.164
H-Index - 152
eISSN - 1466-8238
pISSN - 1466-822X
DOI - 10.1111/j.1466-8238.2012.00760.x
Subject(s) - ranking (information retrieval) , abiotic component , prioritization , regression , ecology , selection (genetic algorithm) , environmental science , biodiversity , climate change , physical geography , geography , biology , computer science , statistics , mathematics , machine learning , management science , economics
Aim We play the role of an ice age ecologist (IAE) charged with conserving biodiversity during the climate changes accompanying the last deglaciation. We develop reserve‐selection strategies for the IAE and check them against rankings based on modern data. Location Northern and eastern North America. Methods Three reserve‐selection strategies are developed. (1) Abiotic: the IAE uses no information about species–climate relationships, instead maximizing the climatic and geographic dispersion of reserves. (2) Species distribution models (SDMs): the IAE uses boosted‐regression trees calibrated against pollen data and CCSM3 palaeoclimatic simulations from 21 to 15 ka bp to predict modern taxon distributions, then uses these as input to the Z onation reserve‐ranking program. (3) Rank‐and‐regress: regression models are used to identify climatic predictors of zonation rankings. All strategies are assessed against a Z onation ranking based on modern pollen distributions. Analysis units are ecoregions and grid cells. Results The abiotic strategy has a negative or no correlation between predicted and actual rankings. The SDM‐based strategy fares better, with a significantly positive area‐corrected correlation ( r = 0.474, P < 0.001) between predicted and actual rankings. Predictive ability drops when grid cells are the analysis unit ( r = 0.217, P = 0.058). Predictive ability for the rank‐and‐regress strategy is similar to the SDM results. Main conclusions For the IAE, SDMs improve the predictive ability of reserve‐selection strategies. However, predictive ability is limited overall, probably due to shifted realized niches during past no‐analogue climates, new species interactions as species responded individually to climate change, and other environmental changes not included in the model. Twenty‐first‐century conservation planning also faces these challenges, and is further complicated by other anthropogenic impacts. The IAE's limited success does not preclude the use of climate scenarios and niche‐based SDMs when developing adaptation strategies, but suggests that such tools offer at best only a rough guide to identifying possible areas of future conservation value.