SDMs - using spatial information to supplement biased occurrence data
For many species the only available distribution data are presence-only occurrences from ad hoc collections such as herbarium or museum specimens.
These data often have major sampling biases (e.g. oversampling near urban areas and roads), which makes it difficult to produce reliable species distribution models.
In this SEEC Stats Toolbox seminar, Vernon talks about a method that helps to address this major flaw of most species species distribution models. He shows how one can integrate other spatial information, such as expert range maps, sampling bias estimation and native range distributions to improve predictions of species distributions.
For example code, please see the following papers and associated scripts:
Merow, C., Allen, J.M., Aiello‐Lammens, M. and Silander Jr, J.A., 2016. Improving niche and range estimates with Maxent and point process models by integrating spatially explicit information. Global ecology and biogeography, 25(8), pp.1022-1036.