Since numerical taxonomy is an operational science, the procedure is divided into a number of repeatable steps, allowing the results to be checked at every step.
i) Choice of units to be studied: The first step is to decide what kind of units to study. In numerical taxonomy, the basic unit of study is called the "operational taxonomic unit" (OTU). Thus the OTU can be an individual plant if the taxonomist is studying a single population of plants to find out the range of variations in its characters. Similarly, you may treat an entire population of plants as an OTU if you are studying a single species represented by different populations existing in nature; or the OTU may be different species when genus'is being evaluated. Therefore, in numerical taxonomy, the OTU varies with the material being studied, and this' helps the taxonomists in making an objective study.
ii) Character selection : After selecting the OTU's, it is necessary to select characters by which they are to be classified. By experience, you will lCarr1 that characters which vary greatly amongst the OTU's are clearly more useful in numerical taxonomy; and we know that as many characters as possible may be used.
Preferably a minimum of GO and generally 80 to 100 or more characters arc needed to produce a fairly stable and reliable classification. The selected characters have then to be coded or given some symbol or mark.' There are 2 methods of coding taxonomic information.
a) Binary coding or two-state coding-This is the simplest form of coding adopted in numerical taxonomy where the characters are divided into + and -, or as 1 and 0. The positive characters are recorded as + or 1 and the negative characters as - or 0. It is possible to use this method of coding for all characters studied.
In case a particular character is not present in an OTU being examined, the symbol or code NC is used, indicating that there is no comparison for that characters. However, we find that by using this method of coding, we tend to increase our work because there are large variations in the plant, and very often a single character such as colour of flower can be represented in a wide range We can have white, pink, red, yellow and other colours in roses. If we are to use this data in a binary coding, then we will have to use each colour as a character and it would be coded as + or -, as the case may be.
b) Multi-state coding-An alternative method would be to use multi-state coding where a single character can be coded in a number of states, each being represented by a numerical symbol or code (e.g. 1,2,3,4,5, . . , . .) depending on the range of variation. Thus, if we again look at the colour of the rose flower, we can give different codes to different colours such as white = 1 , pink = 2, red = 3. yellow = 4, and so on. Besides qualitative characters such as colour of flower . type of placentation, etc., multistate coding is also useful of quantitative characters such as plant height, leaf length, leaf breadth, and other characters involving measurements. A code is prepared for the range of variation and appropriate symbols are allotted to each unit in the. range.
The data obtained by scoring the characters in the OTU's ?re then presented in a table as a .data matrix giving the OTU's on one side of the table and the codes for different' characters against each OTU. Thus, if one has studied 25 OTU's and has scored 75 characters from each, the data matrix will contain 25 x 75 = 1875 units of information. his kind of large it of information in the data matrix necessitates the use' of computers to help the taxgnomists. to digest the knowledge quickly. It is also important to. remember that computer programmes are based on mathematical equations and-computer language and the data matrix is essential for this purpose. In addition, 'the . next step in numerical taxonomy is entirely dependent ;on the data matrix.