1973 REANALYSIS zyxwvutsr OF NEST STRUCTURE IN WEAVERS 93 A REANALYSIS OF NEST STRUCTURE I N THE WEAVERS (PLOCEINAE) USING NUMERICAL TAXONOMIC TECHNIQUES GARY D. SCHNELL Received on 4 October 1971 Ploceine finches construct globular nests with an entrance on the side or bottom that may be extended into a tube or spout. The nest fabric is usually strong and pliant, and often woven by the birds using intricate stitching techniques. Crook (1963) studied extensively nests of 62 species of weavers from India, Africa, Madagascar and the Seychelles; although nests of all species were not examined, he believed that all types of nest architecture in the subfamily were represented. After analysing nine variables of nest architecture and construction behaviour, Crook classified the nests in seven phe- netic groups. He has made considerable use of the data in a later paper (1964) on the evolution of social organisation and visual communication in the weavers. Also Hall zy & Moreau (1970) make numerous references to the work in their treatment of African passerines. In this study I have reanalysed his data using various multivariate statistical techniques and have compared and contrasted Crook’s results with my own. From the standpoint of describing phenotypic variation, it is often difficult to evaluate the results of a more or less intuitive phenetic analysis such as Crook’s, since there is the possibility that charac- teristics other than those coded in tables influenced the resulting groupings. Also, it is essentially impossible to determine what weights were assigned to various characters in constructing a phenetic classification intuitively. The multitude of difficulties inherent in grouping multiple objects by eye on the basis of multiple characters will be readily admitted by those having conducted such studies. Given these considerations, one purpose in undertaking this restudy was to see if numerical methods applied to Crook’s tabled characteristics of weaver nests would result in clusters similar to his. If the resulting classifications are different, what insight might we obtain from analysing the new groupings? MATERIALS AND METHODS The 48 OTUs (Operational Taxonomic Units; Sokal & Sneath 1963) used in the study are listed in Table 1. Generic names are abbreviated in the table and throughout as follows: zyxwvutsr A., zyxwvuts Amblyospiza; F., Foudia; M., Malimbus; P., Ploceus; and Pa., Pachyphantes. OTU numbers were arbitrarily assigned to species in the order of Crook’s (1963) Table 1. OTU 42 represents the 15 species of Euplectes, all of which have essentially the same nest architecture and construction behaviour. All characters were coded from the data in Table 1 of Crook (1963); the reader is referred to his extensive descriptions of character states if further clarification is desired. I recorded his data into 38 two-state characters to allow for numerical analysis. Question marks in his table were ignored; when an “ or ” was included in the table, both character- istics were coded as present; and blanks (i.e. dashes in his table) were coded as no comparison (NC; Sokal & Sneath 1963). NCs make up 7.8% (230 of 1,824 entries) of the data matrix. Table 2 lists the characters used and the character-state values assigned. Numerical taxonomic techniques were used with definitions and terminology taken from Sokal & Sneath (1963). Computations were carried out using NT-SYS (Numerical Taxonomy System), a system of multivariate computer programmes developed by F.