The aim of this study has been to prepare a systematic revision of the genera of the family Chrysobalanaceae. At the outset of this research it was apparent that the distinctions between the subgenera and some other groups within the single genus Parinari Aubl. were much greater than the differences between other genera in the family. This is largely because most recent work has been done on a restricted regional basis and generic concepts differ widely in different regions. Most of the earlier workers only had access to incomplete material. for the present study complete material for more than 200 species was assembled. The wood anatomy of species representing all genera except Kostermansia Prance and Hunga Prance was studied. Pollen slides representing all genera were prepared. Seedlings from twenty-six species were also examined. Much useful anatomical information published by other workers has been brought together in this work. Papers on leaf anatomy by Kanduuml;ster (1897); the ovary by Juel (1915), leaf trace anatomy by Morvillez (1918a), and pedicel and floral anatomy by Bonne (1928) have all been of the greatest use. The first author to give the group its present circumscription was Robert Brown (1818) who recognized it as a family. The last author, however, to monograph this group on a world-wide basis was De Candolle, who, in his 'Prodromus' (1825) placed it as the first tribe of his Rosaceae. Subsequent authors hive been approximately equally divided into those who treat it as a family and those who treat it as a tribe or subfamily of the Rosaceae. However, the authors of the most widely used general systems of classification have been unanimous in placing it in the Rosaceae (Bentham andamp; Hooker, 1865; Focke in Engler andamp; Prantl, 1894; Hutchinson 1926, 1959). Focke's is the last work in which all genera are described. Focke recognized the following genera:- Chrysobalanus L., Grangeria Comm. ex Juss., Moquilea Aubl., Lecostemon ["Lecostemion"] Moc. andamp; Sessandeacute; ex DC. and Stylobasium Desf. in the subtribe Chrysobalanineae, and Hirtella L., Couepia Aubl., Parinari Aubl., Acioa Aubl., Angelesia Korth. and Parastemon A. DC. in the subtribe Hirtellineae. Lecostemon and Stylobasium were included with some doubt and Focke suggested that they might be more closely related to Phytolaccaceae. Jubsequent authors have added the genera Afrolicania Mildbr., Geobalanus Small and Magnistipula Engl. At an early stage of this investigation it was found that Stylobasium and Lecostemon differ from all other Chrysobalanaceae in almost all important respects. Focke, and all previous and some subsequent authors, have wrongly identified Lecostemon. In this work it is shown that the true Lecostemon is in fact a Sloanea of the Tiliaceae and that Lecostemon sensu Focke is correctly named Rhabdodendron, a genus which has been variously accommodated in Rutaceae and Phytolaccaceae. The present study has shown that Rhabdodendron is not only distinct from all Chrysobalanaceae in external morphology, wood anatomy and pollen morphology, but also differs from the Rutaceae in these respects. In wood anatomy it was found to be very similar to Phytolaccaceae. Its pollen is somewhat different from that of the Phytolaccaceae but not appreciably different from other members of the Centrospermae. In external morphology Rhabdodendron has many distinctive features most of which occur sporadically in the Centrospermae but not in combination. In view of this it seems preferable to treat it as a unigeneric family related to but distinct from the Phytolaccaceae. A Latin description of this new family is given, but it is realized that further work on its relationship to Phytolaccaceae is necessary before it should be published. Many authors have suggested that Stylobasium does not belong to the Chrysobalanaceae or is an isolated member within it, but only Agardh (1858) described it as a separate family. It ie shown in this study that Stylobasium is utterly different from all Chrysobalanaceae in external morphology, wood anatomy, pollen morphology and floral anatomy. In wood anatomy and pollen particularly there are striking similarities to certain members of the Sapindales, and it is suggested that Agardh's family should be recognized and placed near Sapindaceae and Anacardiaceae. Purged of these two anomalous genera, the Chrysobalanaceae is now a homogeneous entity, whose wood structure and pollen morphology is so uniform that few genera can be discriminated on the basis of these characters. However, wood anatomy and pollen morphology are found to differ constantly and to an appreciable degree from the Rosaceae so much so that, taken in conjunction with the anatomical features described by Kanduuml;ster, Juel, Morvillez and Bonne, they seem to justify the recognition of the group as a family distinct from, although related to, the Rosaceae. Most previous authors have variously subdivided the group. Their views are briefly summarised, and it is shown that anatomical characters provide no basis for a rational subdivision. In this work, for convenience, two tribes are recognized based on the symmetry of the flower. In the Chrysobalaneae the ovary is inserted at or near the base of the receptacle-tube. In the Hirtelleae the ovary is inserted laterally or at the mouth of the receptacle-tube. Parinari is unique within the family in having its carpels partitioned by a false septum. This character has been used to define Parinari since it was originally described by Aublet in 1775, but visual inspection is enough to show that its uncritical use has given rise to an extremely heterogeneous assemblage. Some components of this are more closely related to genera outside Parinari than to the rest of Parinari. Some species have been assigned to Parinari which do not even nave its artificial unifying feature. It was quite clear that currently accepted generic limits were untenable and that there were two alternative taxonomic procedures. Either all species within the family should be united to form a single genus Chrysobalanus or an attempt should be made to discover more natural groupings. After a detailed study of the external morphology of more than 200 species, the author was satisfied that various segregates of Parinari should be recognized as genera, and that most of the other genera in the Chrysobalanaceae could conveniently be kept apart. However, it was decided to use a computer to demonstrate as objectively as possible the exact correlation of those characters believed by the author to be of greatest taxonomic worth and of all other important characters used by previous authors. For the tribe Hirtelleae (which includes Parinari sens. lat.) eleven qualitative and ten quantitative characters were used and scored numerically for 124 species. An association-analysis was made for the qualitative data and a principle-component analysis for the quantitative data using programmes devised by Professor W.T. Williams and his associates for a Feranti 'Pegasus' computer. The entire data was analysed by a principle-component analysis programme by Mr. J.N.R. Jeffers for a Feranti 'Sirius' computer. This is possibly the first application of these techniques to a problem concerning generic identities of higher organisms. Although similar methods have been used in discriminating between closely related species, they do not seem to have been used at a higher level.
Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:644608 |
Date | January 1963 |
Creators | Prance, Ghillean T. |
Publisher | University of Oxford |
Source Sets | Ethos UK |
Detected Language | English |
Type | Electronic Thesis or Dissertation |
Source | http://ora.ox.ac.uk/objects/uuid:25003fc3-26fe-4cd6-9b90-92f1bd6d31ad |
Page generated in 0.0037 seconds