species.richness.cv {sperich}

Species richness cross-validation

Description

This function cross-validates the species richness based on given species occurrences.

Usage

species.richness.cv(dataset.all.species, landwatermask, fold=5, 
	loocv.limit=10, distances=3:10, weight=0.5, dimension, origin, 
	resolution=1, upperbound, all.species=-1, 
	silent=TRUE, do.parallel=FALSE)

Arguments

dataset.all.species

A dataset containing the species with their ID (named: speciesID)and the longitude (named: long) and latitude (named: lat) of their occurrence location.

landwatermask

A grid containing the land-water-information of the observed area. If a grid cell containes no land, the value of the cell in the landwatermask is -1, otherwise it is 0. Additional, height-informations could be added for land surfaces. In this case, take care of the 'upperbound' value.

fold

The number of groups which should be created if the number of occurrences is greater than loocv.limit.

loocv.limit

The limit below which the subsamples are created for a leave-one-out-cross-validation instead of a k-fold-cross-validation.

distances

The distances which will be used for species range estimation.

weight

The tuning parameter of the weighting procedure (details in Raedig et al. 2010).

dimension

The dimension of the processed grid.

origin

The geographic coordinates of the origin of the grid.

resolution

The resolution of the grid in (geographical) degree.

upperbound

This value determines the height (based on values in 'landwatermask') which is considered to be a barrier for species distribution.

all.species

The vector with the numbers of the species which should be mentioned. If the first value is -1, all species in the database will be used for species richness estimation.

silent

A boolean flag that determines wether the report of status messages should be suppressed or not.

do.parallel

A boolean flag determining wether the function uses the 'foreach'-package to cross-validate via parallel processing.

Details

This routine cross-validates the species richness based on given species occurrences through a geometric interpolation model (details in Raedig et al. 2010).

Value

This function returns a grid which contains the cross-validated species richness information.

Author(s)

Maximilian Lange, Sven Lautenbach, Claudia Raedig

References

Raedig, C., Dorman, C.F., Hildebrandt, A. and Lautenbach, S. (2010). Reassessing Neotropical angiosperm distribution patterns based on monographic data: a geometric interpolation approach. Biodivers Conserv, 19, 1523-1546.

Examples

##load data
data(dataset.all.species)
data(dataset.landwater)

##create grid parameters
dimension <- getDimension(dataset.all.species, resolution=1)
origin <- getOrigin(dataset.all.species)

##create landwatermask
landwatermask.nocoast <- createLandwatermask(dataset.landwater, 
				dimension, origin, resolution=1)

##estimate species richness
species.richness.weighted <- species.richness.cv(dataset.all.species, 
		landwatermask.nocoast, fold=5, loocv.limit=10, distances=2:5, 
		weight=0.5, dimension, origin, resolution=1, upperbound=5, 
		all.species=1:2)