Probing Multivariate Data with Random Skewers: A Preliminary to Direct Gradient Analysis
- 1 January 1984
- Vol. 42 (2) , 161-165
- https://doi.org/10.2307/3544788
Abstract
Suppose an s-species, ecological community is suspected of being influenced by an environmental gradient, and suppose the quantities (or proportions) of the species were observed in a series of n sampling units ranged along the gradient. The observational data constitute (conceptually) a swarm of n points, representing the sampling units and labeled with their rank in the series, scattered in an s-dimensional coordinate frame. One can test whether the pattern of the data swarm reflects the ordering of the sampling units along the gradient by probing the swarm with numerous randomly oriented lines (skewers) and observing the ordering along them of the ranked data points. The test is distribution-free (nonparametric) and is applied to the raw, uncondensed data. It enables one to judge objectively whether there is a significant trend in community structure along the gradient. Application of the test is illustrated with data on the benthic fauna of a river.This publication has 0 references indexed in Scilit: