English   español  
Please use this identifier to cite or link to this item: http://hdl.handle.net/10261/170440
Share/Impact:
Statistics
logo share SHARE   Add this article to your Mendeley library MendeleyBASE

Visualizar otros formatos: MARC | Dublin Core | RDF | ORE | MODS | METS | DIDL
Exportar a otros formatos:
Title

Background sampling and transferability of species distribution model ensembles under climate change

AuthorsIturbide, Maialen; Bedia, Joaquín ; Gutiérrez, José M.
KeywordsPseudo-absences
Quercus robur
Future projections
Variance partitioning
Overfitting
Peripheral populations
Issue Date2018
PublisherElsevier
CitationGlobal and Planetary Change 166: 19-29 (2018)
AbstractSpecies Distribution Models (SDMs) constitute an important tool to assist decision-making in environmental conservation and planning. A popular application of these models is the projection of species distributions under climate change conditions. Yet there are still a range of methodological SDM factors which limit the transferability of these models, contributing significantly to the overall uncertainty of the resulting projections. An important source of uncertainty often neglected in climate change studies comes from the use of background data (a.k.a. pseudo-absences) for model calibration. Here, we study the sensitivity to pseudo-absence sampling as a determinant factor for SDM stability and transferability under climate change conditions, focusing on European wide projections of Quercus robur as an illustrative case study. We explore the uncertainty in future projections derived from ten pseudo-absence realizations and three popular SDMs (GLM, Random Forest and MARS). The contribution of the pseudo-absence realization to the uncertainty was higher in peripheral regions and clearly differed among the tested SDMs in the whole study domain, being MARS the most sensitive — with projections differing up to a 40% for different realizations — and GLM the most stable. As a result we conclude that parsimonious SDMs are preferable in this context, avoiding complex methods (such as MARS) which may exhibit poor model transferability. Accounting for this new source of SDM-dependent uncertainty is crucial when forming multi-model ensembles to undertake climate change projections.
Publisher version (URL)https://doi.org/10.1016/j.gloplacha.2018.03.008
URIhttp://hdl.handle.net/10261/170440
Identifiersdoi: 10.1016/j.gloplacha.2018.03.008
issn: 0921-8181
e-issn: 1872-6364
Appears in Collections:(IFCA) Artículos
Files in This Item:
File Description SizeFormat 
Backchang.pdf Embargoed until July 1, 20203,78 MBAdobe PDFThumbnail
View/Open    Request a copy
Show full item record
Review this work
 


WARNING: Items in Digital.CSIC are protected by copyright, with all rights reserved, unless otherwise indicated.