MetaLandSim (version 1.0.4)

range_raster: Probability of occupancy, dispersal model

Description

This function intends to create a raster map, estimating probability of occupancy, at a given time step, based on species dispersal and landscape configuration. range_raster uses the output from range_expansion and a raster map with the species current occupancy.

Usage

range_raster(presences.map, re.out, mask.map=NULL, plot.directions=TRUE)

Arguments

presences.map

string of the raster file name with species occurrence.

re.out

object of class list expansion. Output from range_expansion.

mask.map

default NULL. String of the raster file name with the mask. Usually, 1 over the area where the analyses should be done.

plot.directions

default TRUE. Whether It will (TRUE) or will not (FALSE) return a graphics for the expansion model functions and raster maps with expansion probabilities in all four cardinal points.

Value

Produces the spatial realization of the dispersal model, composed by a stack of two objects of the class RasterLayer (see Raster-class package for further description), with the probability of occupancy and the time step a given distance is occupied. This version of MetaLandSim uses GRASS, version 7 through the package rgrass7. Additionally these rasters are saved in the working directory (files 'PROB' and 'TSTEP') defined by the user and can be directly imported to any GIS software.

Details

The function automatically reads the raster input files (presences.map and mask.map, if present). Usually, 0 for absence and 1 for presence in every square cell over a given resolution. Supported file types are those that can be read via rgdal (see gdal. Note that the projection for the raster layer should be one of those supporting metric units (i.e., linear scale is equal in all directions around any point such as Transverse Mercator; see http://spatialreference.org/).

Then, it computes and fits single sigmoidal functions for every direction on the expansion movements (building four sub-models, one to each main cardinal direction), as previously computed by range_expansion. Four different raster maps are generated (the sub-models), each estimating the probability of expansion for north, south, east and west directions. The four maps are finally summarized into a single range expansion map, which is returned to the user as an object of class RasterLayer and saved in the working directory. These four maps do not directly express the probabilities in the output of range_expansion. Rather, the outputs are weighted by directionality, so that e.g. the north model favours the dispersal towards the north while penalizing dispersal in every other direction. As such, the resulting dispersal model is not a direct spatial transcription of the four data frames provided by range_expansion but an interpretation weighted by the spatial context given by directionality.

Additionally, a raster file is computed, showing the time steps at which each distance is reached. This output depicts the adjustment of a linear model to the output of the four sub-models together, considering the four equally. This output should not be used as guideline to mask the model if running to several time periods. For example if projections to 2050 and 2080 are to be made than the dispersal model should be run twice, adjusting the time steps to the desired date.

This function internally uses a connection to GRASS GIS software through the package rgrass7-package, in order to increase the performance for geographical calculations.

Finally, the user might have to manually adjust the starting values of the function fit.sigmoid, (defined internally in this function) if it has difficulty adjusting to the output of range_expansion.

References

The same as range_expansion.

See Also

range_expansion, Raster-class, rgrass7, initGRASS

Examples

# NOT RUN {
# }
# NOT RUN {
data(rg_exp)

presences <- system.file("examples/presences.asc", package="MetaLandSim")
mask <- system.file("examples/landmask.asc", package="MetaLandSim")

require(rgrass7)

#Initializing a GRASS session in a temporal directory:
#### Under Linux systems:
initGRASS("/usr/bin/grass", home=tempdir())
#### Under Windows systems:
initGRASS("C:/GRASS", home=tempdir())

range_raster(presences.map=presences, re.out=rg_exp, mask.map=mask) -> range.map

plot(range.map)

require(rasterVis)

levelplot(range.map, contour=TRUE)
# }