Differentiating between distribution and suitable habitat in ecological niche models: A red spruce (Picea rubens) case study.

• There is often a difference between the distribution of a species and the suitable habitat available to it. • Distribution can and should be modeled separately from suitable habitat. • Presence-proximal true-absence points can be used to model distribution as opposed to suitable habitat. • Modeling distribution and suitable habitat simultaneously enables identification and quantification of the environment suitable for restoration management. Ecological niche modeling has become common practice amongst ecologists in the last two decades. The terms in the field of ecological niche modeling, such as ecological niche model, species distribution model, and habitat suitability model are often used interchangeably, although each term can have very different meanings depending on the context of the study. Without full understanding of model inputs and careful interpretation of model outputs, there is a possibility that the dimension of the ecological niche being predicted does not align with the modeling goals of the author. This study showcases an example which indicates that it is possible to intentionally model species distribution and habitat suitability separately. Red spruce (Picea rubens) distribution was modeled by collecting true-absence points in proximity to red spruce presence points. Red spruce suitable habitat was modeled by excluding true-absence points and only using pseudo-absence points along with presence points. Models which included presence-proximal true-absences had functionally greater discriminatory ability in map projections and were more sensitive to subtle environmental changes when compared to models which only included pseudo-absence points. Therefore, models which included true-absence points more closely approximated the current distribution of red spruce whereas models which only included pseudo-absences more closely approximated the potential distribution within the realized niche or suitable habitat of red spruce. This study demonstrates that distribution and suitable habitat can and should be modeled separately and emphasizes the necessity of carefully considering modeling objectives and methodologies when interpreting modeling outcomes. [ABSTRACT FROM AUTHOR]