A Numerical Taxonomic Approach to the Study of Angiosperm Leaves

Abstract

Dolph proposed data banking of angiosperm leaf information. For a data bank to function satisfactorily, the scored character set must be suitable for statistical testing. Such a character set, based on a ratio scale, was compiled from various sources. Using 100 leaves from 20 spp. [Argyrodendron actinophyllum (C. Moore) Edlin; Elaeocarpus grandis F. vM; Castanospermum australe A. Cunn and Fraser; Tristania conferta R. Br.; Flindersia australis R. Br.; Hymenosporum flavum F. vM; Brachychiton discolor F. vM; Harpullia pendula Planch; Linociera ramiflora (Roxb.) Wall.; Toona australis Harms.; Hardenbergia violacea (Schneev.) Stearn; Macropiper excelsum Miq.; Pittosporum undulatum Vent.; Rhodosphaera rhodanthema Engl.; Cinnamomum camphora T. Nees and Eberm; Pittosporum rhombifolium A. Cunn.; Alectryon excelsus Gaert.; Cupaniopsis anacardioides Radlk.; Bauerella australiana Borzi; Laurus nobilis L.], the numerical taxonomic methods clearly clustered the leaves into their species. Several factors, such as character number, character correlation, missing data and sample size and their effect on classification were considered. To increase accuracy of character scoring, a method to determine vein orders was devised. Binary data were also considered and a strategy to describe fossil floras presented.