8th to 12th January, 2015 - Bayreuth, Germany

Community level models outperform traditional species distribution models in no-analog climatic and ecological regimes 

Kaitlin Clare Maguire1, Diego Nieto-Lugilde2, Jessica Blois1, Matthew Fitzpatrick2, John Williams3
1 University of California, Merced
2 Appalachian Laboratory, University of Maryland Center for Environmental Science
3 University of Wisconsin, Madison

P6.10⚡ in Modelling Species and Ecosystems

Species distribution models (SDM) utilize species-environment relationships to predict species response to climatic changes; however, a species response is also influenced by biotic interactions. Recently developed community level models (CLMs) incorporate species associations (i.e. biotic interactions) into models of species responses by using co-occurrence matrices. It is unclear from previous work whether CLMs outperform SDMs when building and projecting models in climatically similar scenarios and it is unknown how the two compare when projected to climatically novel regimes. Here we compare the performance of five SDMs and their CLM equivalents across spatial and temporal scales by capitalizing on the rich fossil pollen record of the Eastern United States through the last 21,000 years. Model performance of both SDMs and CLMs decreases when projected to climatically and ecologically distant time periods but CLMs outperform SDMs in these instances. These results indicate the importance of including biotic interactions into models that predict how species will respond to the novel climatic regimes of the future. 

Keywords: biotic interactions, paleoecology, paleobiogeography, pollen