nodespec: Calculates the node-based specialisation index

In bipartite: Visualising Bipartite Networks and Calculating Some (Ecological) Indices

Calculates the node-based specialisation index

Description

Calculates a specialisation index based on the node positions for all species in a bipartite network, separately for the higher and lower trophic level.

Usage

nodespec(web, inf.replace = NA)

Arguments

web	A matrix with lower trophic level species as rows, higher trophic level species as columns and number of interactions as entries.
inf.replace	What should infinite geodesic distances (e.g. between compartments) be represented as? Defaults to NA; only currently implemented alternative is inf.replace=Inf, which replaces infinite distances by the maximum path length plus 1.

Details

This index aims to describe the functional specialisation of pollinators and was proposed by Dalgaard et al. (2008). It is a purely qualitative measure.

After calculating the geodesic distances between species, i.e. the minimum number of steps from one species to another, these values are averaged for each species. This mean geodesic distance is interpreted as functional specialisation (Dalgaard et al. 2008).

Notice that this “new” index is in fact little else than the inverse of (unscaled) closeness centrality in disguise.

Value

A list with two components, names “higher” and “lower”, both containing the node specialisation index for each species.

Note

This index is as yet unevaluated. We don't know how it responds to true specialisation at all. In fact, it is a rather good example of how to get a new thing published without even having demonstrated in which way it differs from existing indices of specialisation (such as standardised d included in the function dfun), or how it performs on artificial data with known properties.

One major disadvantage of any index based on path lenghts is its difficulty with compartments, i.e. species not linked to the rest of the network. There are, generally speaking, three ways to handle this: Firstly, ignore it (that is, set infinite distances to NA; our default). Secondly, leave it as it is (that is, leave infinite distances as infinite). This is not really an option, since then ALL species would have infinite specialisation values. Thirdly, replace infinite by the largest distance plus one (see comments in geodist in sna). That would probably be a plausible thing to do, since we could argue that with a little bit extra observation we might have found a species linking a compartment to the rest of the network. However, this solution is “not canonical”, as put in geodist and hence biased to an unknown extent. To use this option, specify inf.replace=Inf.

Author(s)

Carsten F. Dormann [e-mail]

References

Dalsgaard, B., Martín González, A. M., Olesen, J. M., Timmermann, A., Andersen, L. H. and Ollerton, J. (2008) Pollination networks and functional specialization: a test using Lesser Antillean plant-hummingbird assemblages. Oikos 117, 789–793

See Also

See also as specieslevel, which calls nodespec.

Examples

data(Safariland)
nodespec(Safariland, inf.replace=Inf)

bipartite documentation built on May 29, 2024, 2:23 a.m.