An Objective Approach to Select Climate Scenarios when Projecting Species Distribution under Climate Change
Open Access
- 25 March 2016
- journal article
- research article
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 11 (3) , e0152495
Abstract
An impressive number of new climate change scenarios have recently become available to assess the ecological impacts of climate change. Among these impacts, shifts in species range analyzed with species distribution models are the most widely studied. Whereas it is widely recognized that the uncertainty in future climatic conditions must be taken into account in impact studies, many assessments of species range shifts still rely on just a few climate change scenarios, often selected arbitrarily. We describe a method to select objectively a subset of climate change scenarios among a large ensemble of available ones. Our k-means clustering approach reduces the number of climate change scenarios needed to project species distributions, while retaining the coverage of uncertainty in future climate conditions. We first show, for three biologically-relevant climatic variables, that a reduced number of six climate change scenarios generates average climatic conditions very close to those obtained from a set of 27 scenarios available before reduction. A case study on potential gains and losses of habitat by three northeastern American tree species shows that potential future species distributions projected from the selected six climate change scenarios are very similar to those obtained from the full set of 27, although with some spatial discrepancies at the edges of species distributions. In contrast, projections based on just a few climate models vary strongly according to the initial choice of climate models. We give clear guidance on how to reduce the number of climate change scenarios while retaining the central tendencies and coverage of uncertainty in future climatic conditions. This should be particularly useful during future climate change impact studies as more than twice as many climate models were reported in the fifth assessment report of IPCC compared to the previous one.Keywords
This publication has 40 references indexed in Scilit:
- Consequences of climate change on the tree of life in EuropeNature, 2011
- The CC-Bio Project: Studying the Effects of Climate Change on Quebec BiodiversityDiversity, 2010
- Bioclimate envelope model predictions for natural resource management: dealing with uncertaintyJournal of Applied Ecology, 2010
- Niches, models, and climate change: Assessing the assumptions and uncertaintiesProceedings of the National Academy of Sciences, 2009
- THE WCRP CMIP3 Multimodel Dataset: A New Era in Climate Change ResearchBulletin of the American Meteorological Society, 2007
- Newer Classification and Regression Tree Techniques: Bagging and Random Forests for Ecological PredictionEcosystems, 2006
- Reducing uncertainty in projections of extinction risk from climate changeGlobal Ecology and Biogeography, 2005
- Predicting the impacts of climate change on the distribution of species: are bioclimate envelope models useful?Global Ecology and Biogeography, 2003
- A review of methods for the assessment of prediction errors in conservation presence/absence modelsEnvironmental Conservation, 1997
- Hierarchical Grouping to Optimize an Objective FunctionJournal of the American Statistical Association, 1963