metaRange: A framework to build mechanistic range models

Mechanistic or process-based models offer great insights into the range dynamics of species facing non-equilibrium conditions, such as climate and land-use changes or invasive species. Their consideration of underlying mechanisms relaxes the species-environment equilibrium assumed by correlative approaches, while also generating conservation-relevant indicators, such as range-wide abundance time-series and migration rates if demographically explicit. However, the computational complexity of mechanistic models limits their development and applicability to large spatiotemporal extents.

We developed the R package ‘metaRange’ that is a modular framework to build population-based and metabolically constrained range models. It provides a function catalogue for users to calculate niche-based suitability, metabolic scaling, population dynamics, biotic interactions, and kernel-based dispersal, which may include directed movement. The framework’s modularity enables the user to combine, extend, or replace these functions, making it possible to customize the model to the ecology of the study system. The package supports an unlimited number of static or dynamic environmental factors as input, including climate and land use.

As an example, we simulated 100 virtual species in Germany on a 1 km2 resolution over 110 years under realistic environmental fluctuations in three scenarios: without competition, with competition, and with competition and a generalist-specialist trade-off. The results are in accordance with theoretical expectations. Due to the population-level, the package can execute such extensive simulation experiments on regular enduser hardware in a short amount of time. We provide detailed technical documentation, both for the individual functions in the package as well as instructions on how to set up different types of model structures and experimental designs.

The metaRange framework enables process-based simulations of range dynamic of multiple interacting species on a high resolution and low computational demand. We believe that it allows for theoretical insights and hypotheses testing about future range dynamics of real-world species, which may better support conservation policies targeting biodiversity loss mitigation.