Inspiration for this post Conservation is often the main motivation behind studying where a species lives – having a model of a species’ range can help scientists assess whether it is at risk of extinction, designate protected regions to preserve its habitat, and study potential impacts of human activity. When we create species distribution models using common methods like Maxent, the result is a map of predicted habitat suitability or probability of species presence, such as the one below.
Inspiration for this post In species distribution modeling, one of the key steps requires the researcher to select a “background region” for the species, i.e. a region over which a machine learning model will compare the environment of the “background points” with the environment at points where the species is known to occur. The key to selecting this region is to pick an area where the species could occur but hasn’t necessarily been observed – for example, you don’t want to include an area separated from the rest of the range by a big mountain range that you don’t believe the organism could cross, but you do want to include a range of potential environments.