make.disp.prob.v2: Generate dispersal probabilities

In akeyel/spatialdemography: A Spatially Explicit Metacommunity Model

Description

Calculates dispersal probabilities (point to area) via simulation based on It creates a table with dispersal probabilities to different cells compatible with the SpatialDemography model. an input mean dispersal distance (not log transformed) and dispersal standard deviation (log-transformed).

Usage

make.disp.prob.v2(funct, par1, par2, outpath, num.recs = 1e+05,
  max.disp.extent = 101, cut.off = 1e-05)

Arguments

funct	Chose a function for modeling dispersal (e.g., log-normal). Function codes are 1: log-normal, 2: normal, 3: constant, 4: uniform, 5: uniform, rounded to nearest integer.
par1	First function parameter. For log-normal, this should be the mean dispersal distance in cell units. For example, if mean dispersal for a species was 25 m, and you intend to use a 50 m cell size, you would enter 25 / 50 = 0.5 here. Note that this is NOT log-transformed
par2	Second function parameter. For log-normal, this is the dispersal standard deviation in log-transformed cell units.
outpath	outpath specifies where to put the newly generated dispersal probability table
num.recs	The number of draws used in the simulation. Ideally this will be a large number (e.g., 1,000,000), but the function will run slowly for numbers >10,000.
max.disp.extent	The maximum dispersal distance allowed for calculation of probabilities. With the log-normal function and a high enough standard deviation, you can get VERY long distance dispersal. This option sets a cut-point to improve computational efficiency.
cut.off	The minimum allowed dispersal probability. Probabilities less than this value will be rounded to 0. This is intended to improve computational efficiency.

Details

Point to area dispersal assumes that dispersers originate in the center of the cell, and determines dispersal probabilities using the entire area of potential destination cells. (see Chipperfield et al. 2011 for validation of this approach). Unlike Chipperfield et al. 2011, this function is not based on integrals, but uses a simulation approach to approximate the dispersal probabilities.(not as mathematically sophisticated, but much simpler to program, and with large enough sample sizes, the differences are minimal.)

Value

No value is returned, the function is run for its side-effects (creation of a dispersal probability table.)

Note

WARNING: If dispersers disperse out of the landscape extent (set by max.disp.extent), they will still disperse out of the landscape, regardless of whether a torus or absorbing edge is used. Only tested for log-normal distribution so far. Also, it may be faster to use a GIS implementation for this process.

Author(s)

Alexander "Sasha" Keyel

References

Chipperfield, J.D., E.P. Holland, C. Dytham, C.D. Thomas, T. Hovestadt. 2011. On the approximation of continuous dispersal kernels in discrete-space models. Methods in Ecology and Evolution 2: 668-681.

akeyel/spatialdemography documentation built on May 12, 2019, 4:43 a.m.