Species distribution modeling

Species distribution modeling uses observed occurrences and environmental data to estimate or predict where a species can occur. Models range from statistical correlations to mechanistic representations of environmental tolerances.

Ecological niche: The set of environmental conditions and resources that permits a species to persist and reproduce. Models commonly estimate the environmental niche associated with recorded occurrences.
Species occurrence record: A documented observation that a species was present at a particular place and time. These records anchor the geographic and environmental patterns fitted by distribution models.
Species distribution map: A geographic representation of a species' observed or modeled occurrence, suitability, or probability of presence. Maps communicate model outputs, while their meaning depends on the mapped quantity and threshold.
Ecological niche modeling: The modeling of environmental conditions associated with a species to characterize its ecological niche. The term often emphasizes environmental relationships, while distribution modeling emphasizes geographic predictions.
Maximum entropy: A modeling principle that selects the least-biased probability distribution consistent with specified constraints. MaxEnt applies this principle to estimate relative suitability from presence records and environmental predictors.
Environmental covariate: A measured environmental variable used to explain or predict variation in a biological response. Climate, terrain, and land-cover covariates provide predictors for estimated distributions.
Climate change refugium: An area where environmental conditions remain comparatively favorable for a species as climate changes. Projections can identify candidate refugia where future climate may remain suitable.
Habitat suitability modeling: The estimation of how well locations provide habitat conditions for a species or ecological group. It focuses on habitat quality, which is not identical to predicting where a species occurs.
Generalized linear model: A statistical model relating a response variable to predictors through a link function and an error distribution. It can estimate occurrence probability from environmental covariates when presence and absence data are available.
Absence data: Observations establishing that a species was not detected at a specified place and time. Reliable absences distinguish true nonoccurrence from locations that were simply not surveyed.

Linked from 15 pages