rarefied_convex_hull: Rarefied estimates of n-dimensional convex hull volume

In fruciano/GeometricMorphometricsMix: Miscellaneous functions useful for geometric morphometrics

Rarefied estimates of n-dimensional convex hull volume

Description

Computes rarefied estimates of convex hull volume (i.e., the n-dimensional volume of the convex hull containing a set of observations).

Usage

rarefied_convex_hull(Data, rep = 100, samplesize)

Arguments

Data	Matrix or data frame containing data (observations in rows, variables in columns).
rep	number of resamplings to obtain the rarefied estimate
samplesize	sample size to which the rarefaction procedure is carried out

Details

The advantage of computing rarefied estimates of a parameter is to account for differences in sample sizes. This can be useful to compare estimates among groups.

n-dimensional convex hulls are not frequently used in geometric morphometrics to quantify morphospace occupation (but see Drake & Klingenberg 2010; Fruciano et al. 2012, 2014). However, they can be useful to quantify the spread of the points (observations, individuals) in multivariate (tangent shape) space. Because of their own nature, they are obviously sensitive to outliers. For this reason, rarefaction - as implemented here - is particularly useful to compare patterns across groups.

Value

The function outputs a list with the following elements:

RarefiedSamplesEstimates

Estimates of both multivariate variance and mean pairwise Euclidean distance for each of the rarefied samples

Descriptives

Mean, standard deviation, median and 2.5% (Min) and 97.5% (Max) across the rarefied samples

Notice

The function uses internally the package geometry, which should be installed prior to using this function. Notice also that computation of multidimensional convex hulls is computationally expensive and that requires more observations (individuals) than variables. For this reason, errors might be produced when using too small sample sizes and/or too many variables. A solution for this is increasing sample sizes or performing dimensionality reduction by using this approach only on a meaningful subset of principal components. The function does NOT perform GPA on each rarefied sample this may or may not make a difference in estimates. In many cases, it will probably not make much difference.

Citation

If you use this function, in addition to this package, please cite Fruciano et. al (2014), which is the first use of rarefied n-dimensional convex hull in morphometrics.

References

Drake AG, Klingenberg CP. 2010. Large-scale diversification of skull shape in domestic dogs: disparity and modularity. American Naturalist 175:289-301.

Fruciano C, Tigano C, Ferrito V. 2012. Body shape variation and colour change during growth in a protogynous fish. Environmental Biology of Fishes 94:615-622.

Fruciano C, Pappalardo AM, Tigano C, Ferrito V. 2014. Phylogeographical relationships of Sicilian brown trout and the effects of genetic introgression on morphospace occupation. Biological Journal of the Linnean Society 112:387-398.

See Also

rarefied_disparity, BTailTest

fruciano/GeometricMorphometricsMix documentation built on Jan. 31, 2024, 6:24 a.m.