Journal Number 93
November 2005


Taxonomy Made Easy
By Ian St George


Morphology (structure, shape) is the main way to classify organisms into taxonomic groups or taxa.

Early systems relied on only a few features (for instance plants were classified into herbs, shrubs,
trees, climbers, on the basis of their habit; Linnaeus suggested a system based mainly on the features of stamens and carpels). Later, as in Bentham and Hooker's system of classification of plants, many morphological features were considered.

Similarities in morphological features are used for grouping plants; differences are used for separating them. Plants with great differences are regarded as unrelated or distantly related.

For instance, all flowering plants with ovules inside the ovary cavity are grouped as Angiosperms,
which are then divided into Dicotyledons and Monocotyledons, on the basis of different features
of the root system, leaf venation, flower symmetry and number of cotyledons in the embryo.

Numerical taxonomy

The major problem with the Linnaean system is that it is subjective. Different people interpret different groups because each level is arbitrarily defined. So, beginning in the 1950s, scientists began looking for new methods of classifying organisms. This gave rise to numerical taxonomy.

Numerical taxonomy is the classification of organisms by mathematical means. It is based on counting observable features of organisms and may be operated at various taxonomic levels to deal with species or higher taxa. It involves the grouping and computation of the similarity of characters; the results are usually displayed graphically, as a phenogram or dendrogram.

The goal of numerical taxonomy was to be objective. This was to be achieved by converting all observations into numbers and then using a predefined calculation to divide organisms into taxa. However, there was still a lot of subjectivity in selecting the features to be counted and in how the observations were converted into numbers. 

Chemical taxonomy

The presence and distribution of various chemical compounds in plants serve as taxonomic evidence.

Phylogenetic considerations

In the more recent systems, greater emphasis is given to the phylogenetic arrangement of plant groups, based on the evolutionary sequence of the groups, and reflecting their genetic similarities.

The principle of phylogenetics is that organisms should be classified the way they evolved. Our present knowledge of the evolutionary history of plant groups is incomplete, so at best modem systems are a judicious combination of morphological and phylogenetic systems.

Today, phylogenetics is most commonly done at a molecular level. A gene (DNA) or protein sequence is chosen based on a number of criteria. This same sequence is then determined for a number of different organisms and all the sequences are aligned to each other using a multiple sequence alignment program. From this alignment, a phylogenetic tree is created from tree building algorithms to show graphically the sequences and how they are related. There are many ways of determining evolutionary relatedness from multiple sequence alignment, including maximum likelihood, maximum parsimony, pairwise distance and others.

Phylogenetics has emerged as a leading taxonomic method. However, there is still controversy as to its validity and reliability. Since evolution is in the past, each step in the process requires certain assumptions. In addition, different methods perform different analyses and come to different conclusions. To make an analysis as valid as possible, the appropriate method must be used with the appropriate data. 

Just after I wrote this, David McConachie sent me this piece by Oliver Sparrow, written for Orchids Digest.

"Any population has a variance associated with it. Plot any two characteristics of a species - petal width, petal length - for several dozen field-measured representatives and you do not get a point, but a blob. The issue is whether a related but potentially distinct blob overlaps or is distinct enough to make it useful to treat its members as distinct. There are three ways into this.

"The first, which I find preferable over the others, is for someone who knows the class of organism well - and these specific populations in particular - to make a judgement as to whether the way they live in the round makes them truly distinct. Essentially, is it helpful to the expert mind to separate these entities or not? And by expert, I don't mean someone adept at whisker-counting, but one possessed of an ecological expertise which asks whether the lives led by the populations makes them effectively distinct in habit, sexual transmission and role.

"The second is to apply rigor to the phenotypes. This uses principal component analysis to arrive at a tree structure.... This procedure removes - or renders formal - the human judgement of what matters.

"The third procedure uses information from the analysis of the genome and matters dependent on it. This is pretty primitive at the moment: one or more genes only, difference measured not for what it says but for how it says it. I suspect this approach will mature as understanding of the proteome evolves: that is, what turns off and on in response to which signals in order to generate a leaf, this kind of leaf, this kind of leaf with hairs....

"There is no universal way of dividing a population into sets , any more than there is one answer to the question `why?' (Why is that flower red? Because Mrs Jones chose it and she likes red; because red sells best so horticulturists breed it and florists stock it; because of anthrocyanin; because humming birds see red; because that is the colour worn by grooms at weddings ...)." 


1. subjects/biology-edited/chap13/b1313501.asp
2. Sparrow, Oliver.




Previous Page

Journal Index

Next Page

 Journal 93