Are species lists derived from modeled species range maps appropriate for macroecological studies? A case study on data from BIEN

Abstract

Because survey-based species lists for fine-scale areas are generally lacking, ecologists have been using species lists derived from modeled range maps in macroecological studies. Here I evaluate whether it is appropriate to use species lists derived from modeled range maps in macroecological studies. I compared species lists derived from about 90,000 range maps in the Botanical Information and Ecology Network (BIEN) for over 180 geographic regions (12 broad regions covering the whole of the New World; 172 smaller regions, i.e. Taxonomic Databases Working Group units) with those representing complete or nearly complete species lists obtained from the literature for the regions. I determined the completeness of native species composition and proportion of false presences and exotic (non-native) species in species lists derived from range maps in BIEN. I determined whether errors in species lists derived from modeled range maps affect macroecological inferences by comparing the relationships of α- and β-diversity with geographic and climatic variables between different datasets (i.e. species lists derived from range maps versus complete or nearly complete species lists derived from the literature). The completeness for native species lists derived from range maps was 60–74%, depending on spatial scale. On average, 65% of the species in a species list derived from BIEN for a geographic area are false presences or non-native species to the area. The relationships of α- and β-diversity derived from the BIEN dataset with geographic and climatic variables differ significantly from those derived from complete species lists. This study shows that the results of a macroecological study based on species lists derived from modeled range maps may be substantially incorrect due to high percentages of missing native species, and high numbers of non-native species and false presences. Thus, it may not be appropriate to use species lists derived from modeled range maps in the macroecological studies that require complete or nearly complete species lists.