Description

Performs detrended correspondence analysis and basic reciprocal averaging or orthogonal correspondence analysis.

Usage

Arguments

Details

In late 1970s, correspondence analysis became the method of choice for ordination in vegetation science, since it seemed better able to cope with non-linear species responses than principal components analysis. However, even correspondence analysis can produce an arc-shaped configuration of a single gradient. Mark Hill developed detrended correspondence analysis to correct two assumed ‘faults’ in correspondence analysis: curvature of straight gradients and packing of sites at the ends of the gradient.

The curvature is removed by replacing the orthogonalization of axes with detrending. In orthogonalization successive axes are made non-correlated, but detrending should remove all systematic dependence between axes. Detrending is performed using a five-segment smoothing window with weights (1,2,3,2,1) on mk segments — which indeed is more robust than the suggested alternative of detrending by polynomials. The packing of sites at the ends of the gradient is undone by rescaling the axes after extraction. After rescaling, the axis is supposed to be scaled by ‘SD’ units, so that the average width of Gaussian species responses is supposed to be one over whole axis. Other innovations were the piecewise linear transformation of species abundances and downweighting of rare species which were regarded to have an unduly high influence on ordination axes.

It seems that detrending actually works by twisting the ordination space, so that the results look non-curved in two-dimensional projections (‘lolly paper effect’). As a result, the points usually have an easily recognized triangular or diamond shaped pattern, obviously an artefact of detrending. Rescaling works differently than commonly presented, too. decorana does not use, or even evaluate, the widths of species responses. Instead, it tries to equalize the weighted variance of species scores on axis segments (parameter mk has only a small effect, since decorana finds the segment number from the current estimate of axis length). This equalizes response widths only for the idealized species packing model, where all species initially have unit width responses and equally spaced modes.

The summary method prints the ordination scores, possible prior weights used in downweighting, and the marginal totals after applying these weights. The plot method plots species and site scores. Classical decorana scaled the axes so that smallest site score was 0 (and smallest species score was negative), but summary, plot and scores use the true origin, unless origin = FALSE.

In addition to proper eigenvalues, the function also reports ‘decorana values’ in detrended analysis. These ‘decorana values’ are the values that the legacy code of decorana returns as eigenvalues. They are estimated internally during iteration, and it seems that detrending interferes the estimation so that these values are generally too low and have unclear interpretation. Moreover, ‘decorana values’ are estimated before rescaling which will change the eigenvalues. The proper eigenvalues are estimated after extraction of the axes and they are the ratio of biased weighted variances of site and species scores even in detrended and rescaled solutions. The ‘decorana values’ are provided only for the compatibility with legacy software, and they should not be used.

Value

decorana returns an object of class "decorana", which has print, summary and plot methods.

Note

decorana uses the central numerical engine of the original Fortran code (which is in the public domain), or about 1/3 of the original program. I have tried to implement the original behaviour, although a great part of preparatory steps were written in R language, and may differ somewhat from the original code. However, well-known bugs are corrected and strict criteria used (Oksanen & Minchin 1997).

Please note that there really is no need for piecewise transformation or even downweighting within decorana, since there are more powerful and extensive alternatives in R, but these options are included for compliance with the original software. If a different fraction of abundance is needed in downweighting, function downweight must be applied before decorana. Function downweight indeed can be applied prior to correspondence analysis, and so it can be used together with cca, too.

The function finds only four axes: this is not easily changed.

Author(s)

Mark O. Hill wrote the original Fortran code, the R port was by Jari Oksanen.

References

Hill, M.O. and Gauch, H.G. (1980). Detrended correspondence analysis: an improved ordination technique. Vegetatio 42, 47–58.

Oksanen, J. and Minchin, P.R. (1997). Instability of ordination results under changes in input data order: explanations and remedies. Journal of Vegetation Science 8, 447–454.

See Also

For unconstrained ordination, non-metric multidimensional scaling in monoMDS may be more robust (see also metaMDS). Constrained (or ‘canonical’) correspondence analysis can be made with cca. Orthogonal correspondence analysis can be made with corresp, or with decorana or cca, but the scaling of results vary (and the one in decorana corresponds to scaling = "sites" and hill = TRUE in cca.). See predict.decorana for adding new points to an ordination.

Examples