MIAmaxent R package published in Ecology and Evolution

Map of Central and South America from Yucatán to Brazil with different-coloured points over the land area

Presence points of a species in Central and South America. The most common data used in species distribution models are observations that the species is present in a location.

Foto: Vollering, Halvorsen & Mazzoni / Ecology and Evolution

Julien Vollering, Rune Halvorsen and Sabrina Mazzoni's R package and paper has just come out in Ecology and Evolution. The package allows scientists to make more ecologically-informed choices when they build models of species distributions.

Species distribution models are widely used in science, forming a basis for management of the species and a tool to understand their ecology. A favourite software for creating these models is Maxent; it is easy to operate and produces maps that fit the input data well.

The MIAmaxent package introduces a statistical way of modeling species distributions based on Maxent, but with a different type of model selection. Uncritical use of the Maxent software has recieved criticism for making overly-complex models that are not easy to interpret in terms of ecology, making it necessary to take control over the model-building process rather than running Maxent with standard settings.

Making species distribution models with MIAmaxent typically provides simpler models that are ecologically more interpretable. Vollering, Halvorsen and Mazzoni's fundamental motivation for making MIAmaxent was to make distribution models more grounded in ecological theory.

MIAmaxent includes tools for "exploring data and interrogating models in light of knowledge of the modeled system" as well as a logistic-regression alternative when not only presences but also absences of the modelled species are available.

 

Emneord: Distribution modelling, MIAmaxent, MaxEnt, species distributions Av E.L. Eriksen
Publisert 3. okt. 2019 10:22 - Sist endret 26. mars 2024 13:25