A REANALYSIS OF NEST STRUCTURE IN THE WEAVERS (PLOCEINAE) USING NUMERICAL TAXONOMIC TECHNIQUES

Abstract

Summary.Variables of nest architecture and construction behaviour of 62 species of ploceine finches studied by Crook (1963) were reanalysed using multivariate statistical techniques. Phenetic assessments were based on 38 two‐state characters for 48 OTUs. Product moment correlation coefficients and average distance coefficients were used as measures of similarity, and species were clustered using the unweighted pair‐group method with arithmetic averages. A three‐dimensional model was constructed of species plotted with respect to the first three principal components extracted from a matrix of correlations among characters. Principal component and factor analytic techniques were employed to determine trends in character variation.The phenetic affinities expressed in the 3‐D model and two phenograms show many similarities to the phenetic groupings deduced by Crook. However, there were also some differences in interpretation. Crook's Modes C and D were shown as distinct clusters in my analyses, but Modes A and B did not separate completely. In particular, the globular nest of P. baglafecht of Mode A, Type IV, grouped with kidney‐shaped nests of species of Mode B, Type Va. The treatment of P. pelzelni, P. madagascariensis, and P. philippinus in my analysis differed somewhat from Crook's. The resulting cluster of F. sechellarum and M. malimbicus brought out some effects of standardising data on resulting interpretations of similarity.Species within the subtypes of Crook usually remained close together, but the affinities between subtypes (and types) often differed between our studies. The results of factor and principal component analyses indicate that the main trends in character variation concern those traits found together in (a) M. rubriceps; (b) A. unicolor and Pa. superciliosus; (c) forest birds; and (d) non‐forest birds. These groups also are the main clusters. Differences in Crook's and my results are discussed and some of the advantages of utilising numerical techniques to describe similarity are mentioned.