A Generalized Distance Metric for the Analysis of Variable Taxa

Publisher Website
Google Scholar
Abstract
Measures of resemblance are generally inadequate in handling within-operational taxonomic unit (OTU) character variability. A new and generalized Euclidean distance metric is related to existing association coefficients and can be isotonic with commonly applied distance measures. Distortions of the distance matrix arise as a result of distributions of data, properties of special types of characters and intra-OTU variability. The use of equal frequency classes for characters can overcome these problems and validate the assumptions necessary for the probabilistic interpretation of distance values. A conditional binomial probability model is demonstrated and related to the distance metric.

This publication has 3 references indexed in Scilit: