r2dexternal {bipartite}

Generates null models for network analysis by considering external abundances

Description

An extension of r2dtable (and vaznull, respectively) which rescales marginal totals according to independent data

Usage

r2dexternal(N, web, abun.higher=NULL, abun.lower=NULL)

vaznullexternal(N, web, abun.higher=NULL, abun.lower=NULL)

Arguments

N

number of null models to be generated.

web

Web is a matrix representing the interactions observed between higher level species (columns) and lower level species (rows).

abun.higher

Optional vector of externally measured abundances of the higher level. If missing (NULL) it will be replaced by column totals.

abun.lower

Optional vector of externally measured abundances of the lower level. If missing (NULL) it will be replaced by row totals.

Details

The underlying functions are r2dtable and vaznull, which require a vector of row and column totals or a web, respectively. In function nullmodel, these marginal totals are computed from the observed interaction matrix. Here, external abundances can be provided. These will be rescaled and combined with the observed marginal total to construct new row and column vectors for r2dtable.

If neither row nor column abundances are provided this function will be identical to r2dtable and vaznull, respectively.

Value

Returns a list of N null model-generated networks. Species names are (obviously) dropped.

Note

Since the function contains a rounding operation, it also has to include a re-distribution of 1s to keep all species in the system. That means, if one species has been observed in the external abundances extremely rarely (compared to the others), it will be overrepresented in this null model, because otherwise it would have to be dropped altogether! If you have a better solution, please let me know.

When you hand over a web with lots of empty columns/rows, chances are that you have more columns/rows than interactions. In this case you must provide an external abundance vector, otherwise the function will throw an error. If you cannot provide an external abundance vector, consider removing all empty columns and rows (using empty) before applying a null model.

Also note that vaznull itself does not return a matrix with exactly the same marginal totals as the input, so don't expect any different from vaznullexternal!

Author(s)

Carsten F. Dormann [e-mail]

See Also

nullmodel, vaznull, mgen, vegan::simulate and r2dtable

Examples

## Not run: 
	abun.lower <- c(15,5,2,7,4,8,6,0.01,6)
	 set.seed(2)
	(abun.higher <- rpois(27, lambda=4))
	abun.higher[1] <- 0.001
	sum(ext.polls)
	## note: external abundances do not sum up; this is intentional!!
	r2dexternal(2, Safariland, abun.higher=abun.higher, abun.lower=abun.lower)
	r2dexternal(2, Safariland, abun.higher=abun.higher)

  vaznullexternal(2, Safariland, abun.higher=abun.higher, abun.lower=abun.lower)
	
## End(Not run)