OOS 19-3 - Using museum data for species distribution modeling: The case of plants in Florida

Wednesday, August 10, 2016: 8:40 AM
Grand Floridian Blrm G, Ft Lauderdale Convention Center
Charlotte Germain-Aubrey1, Julie Allen2, Robert Guralnick3, Shawn Laffan4, Brent D. Mishler5, Kurt M. Neubig6, Douglas Soltis3, Lucas Majure7 and Pamela Soltis3, (1)University of Florida, Gainesville, FL, (2)Illnois Natural History Survey, Univertisty of Illinois, Champaign, IL, (3)Florida Museum of Natural History, University of Florida, Gainesville, FL, (4)School of Biological, Earth and Environmental Science, University of New South Wales, Australia, (5)Integrative Biology, University of California, Berkeley, Berkeley, CA, (6)Department of Plant Biology, Southern Illinois University, Carbondale, IL, (7)Desert Botanical Garden
Background/Question/Methods

iDigBio is aggregating a volume of digitized data that will allow us to answer questions at unprecedented scales. Florida is home to over 4,100 species of plants and several biodiversity hotspots. We used the historical data linked with herbarium specimens and constructed a pipeline that extracts environmental variables from the time of collection for 1,507 species. 

Results/Conclusions

Also, in view of over- and under-estimations of the actual distribution of the species, we have developed a pipeline to limit the geographical range that the model is built in. Comparing the models with the ranges from experts, we are refining the models and find that a combination of models from several methods is the best approach to obtaining a coherent ensemble of species distribution estimations. This approach is powerful as it can be applied to areas where expert knowledge, or its accessibility, is limited. Moreover, we reconstructed the regional phylogeny for those species to calculate phylogenetic diversity, and include uncertainty from the tree into the phylogenetic diversity measures. This study emphasizes the importance of museum collections to predict future changes in biodiversity at the landscape level.