The geographic distribution is a result of ecological, evolutive and historical process. One way to represent such aspects is by BAM scheme, a heuristic framework that states that the occupied area of a species occurs on sites that have been accessible by means of dispersal (M) and have both favorable biotic (B) and abiotic conditions (A). Although recent years have seen the development of hybrid dynamical models to estimate species distributions, most efforts fail to provide theoretical insights. Here we present the 'bam' software, an R package with tools for modeling dynamically geographic distributions on the basis of the BAM scheme.
The model behind the package is a cellular automaton, with binary states governed by the interaction of movements and biotic and abiotic suitability, represented as a matrix equation. This can be studied both by simulation and using matrix properties (spectral and others). Since realistic matrices are very large, we resort to sparse matrix operations, leading to a practical software package. To illustrate results we use Dismorphia amphione (Lepidoptera) in Mexico, with a spatial grid ~18 km2 of resolution, and using different dispersal scenarios.
Results/Conclusions
Using the adjacency list of the product of suitabilities and movements We found that Dismorphia amphione needs to travel ~160km to occupy its whole potential area of distribution. A scenario where the species can travel 18km leads to eleven non-connected areas. These results illustrate the use of the tool to complement conventional niche models with dispersal hypothesis to display clusters of connected suitable grid-cells, as a function of the dispersal capabilities of a species. To predict the actual path of movements, the full simulation is needed. We illustrate contrasting patterns of invasion for different initial conditions, and calculate the time required for full occupation of the suitable areas.
Although the main objective of the presentation is to show the package functionalities, results demonstrate possible applications of it, for example, the estimation of number of non-connected areas, given dispersal capabilities of a species; the routes that the species needs to take in order to arrive to a certain location; and the size of connected regions. Finally, it is worth noting that the package allows the user to estimate the dispersal dynamics of over tens of thousands of grid cells in just few minutes.