The widely distributed earthworm genus Pheretima auct. containing 746 nominal species and subspecies was last reviewed in 1900 when 167 species were listed. In an attempt to increase our understanding of this group of earthworms numerical methods were used to investigate the affinities of 114 species assigned to the genus with the view to dividing it into smaller, more manageable units. A similarity matrix was obtained from a consideration of 56 characters, it was examined by the method of principal co‐ordinates analysis firstly to which a minimum spanning tree was applied, then secondly by using a five nearest neighbours' procedure. From the arrangement of the species, assemblages were detected which permitted the recognition of eight genera (one being divided into subgenera): Archipheretima, Pithemera gen. nov., Ephemitra gen. nov., Metapheretima, Planapheretima, Amynthas, Metapbhe nom. nov., Pheretima (Pheretima) and Pheretima (Parapheretima). Following research into the literature, most known species were assigned to genera and to nominal species‐groups within the genera, only a few being listed as species incertae sedis. Keys are provided to the genera and to the species and nominal species‐groups. Four Appendices are included, one contains details of a small collection of earthworms from Mount Kinabalu, Sabah, made by the Royal Society Expedition to North Borneo, with the descriptions of a new species and a subspecies. Another concerns the identity of Metaphire fordi (Michaelsen, 1934).

SUMMARY ‘Classica’ classifications of Megascolecoid earthworms are reviewed and the assumptions on which they were based are described. Formerly morphological series were regarded as phylogenetic series and member species were believed to be ancestors and descendants living together contemporaneously. ‘Family‐tree’ type classifications were constructed on these sequences and the systems were regarded as ‘natural’, i.e. phylogenetic, classifications. More recently Omodeo (1958), Lee (1959) and Gates (1959) separately rejected the concept of classifying new taxa by definition, that is, placing them in the appropriate places in a family‐tree according to the weighting given to single characters. Instead each proposed a new classification for the Megascolecoid earthworms based on overall similarities. The three classifications are incompatible. Taxonometric methods are described by which a less subjective approach to the problem of Megascolecoid classification is possible by taking into account a relatively large number of attributes. Thirty species, representing twenty‐nine genera of Megascolecoid earthworms were examined. Forty‐three attributes, eighteen quantitative and twenty‐five qualitative, were scored and the assessments computed to produce a matrix of similarity coefficients. The quantitatives were computed to permit comparisons to be made over the total range of the character states assessed (similar to the requirements of Cain & Harrison, 1958), while the qualitatives were treated on a match or mismatch basis. The matrix of similarity coefficients was examined by two methods. Firstly, by clustering by single linkage, ‘Sneath's method’, to produce a dendrogram and secondly, by transforming the original matrix by ‘Gower's formula’ and calculating the latent roots and vectors of which the first and the second were employed as co‐ordinates in a vector diagram. Two assemblages of genera could be detected in the dendrogram but they did not cluster in the vector diagram. Instead they formed a scatter with genera of one assemblage occupying one‐half of the diagram and genera of the other assemblage one quadrant of the remaining half. The fourth quadrant contained a third assemblage comprising genera which, due to their low similarity coefficients, could not be recognized in the dendrogram as having any special mutual similarity. The pattern of arrangement of the genera was found to coincide largely with the classification previously proposed by Gates and to disagree with those proposed by both Omodeo and Lee. It is concluded that: (a) Problems of oligochaete systematics are suited for study by computer techniques, (b) The results of newer taxonometric procedures need not be a variance with the results obtained by more traditional methods, (c) In this particular investigation, the classification proposed for the Megascolecoid earthworms by Gates (1959), has been confirmed. (The ‘key’ or ‘spot’ characters by which the component family groups may be recognized are firstly, the structure of the prostatic glands and ...

No abstract

Summary 1. Three hundred and fifty‐one specimens of Forest, or, Three‐toed Kingfisher from Malaysia and nearby islands were examined in a quantitative study of the characters distinguishing the dark‐backed taxon, Ceyx erithacus, from the rufous‐backed taxon, C. rufidorsus. 2. All the characters were shown to be variable and some form of intergradation between homologous characters was present in the majority of the sample. 3. The frequency with which intergradation was found to occur between allo‐patric populations indicated that the two taxa freely interbreed wherever they come into contact. 4. It is proposed that the taxa should be regarded as forming one polytypic species, Ceyx erithacus (Linnaeus). 5. In a brief review of the palaeogeography of the area it was suggested that the rufous bird evolved in the islands of Malaysia while secondary isolation in northeastern Borneo resulted in the evolution of the dark‐winged rufous‐backed form characteristic of northern Borneo. 6. With a fall in sea level at the height of the ice‐age there was a resumption of gene flow between previously isolated populations. 7. Currently some of the effect of the exchange of genes is not readily evident due to swamping. 8. A brief systematic review of the races of Ceyx erithacus is included.