Biosystematics of Begonia squamulosa Hook.f. and affiliated species in section Tetraphila A.DC.

This study deals with the systematics of plants belonging to a part of Begonia section Tetraphila that occur in tropical Africa. Six taxa are recognized and accorded the rank of species. The names of three of these taxa, viz.: B. elaeagnifolia Hook.f., B. longipetiolata Gilg and B. squamulosa Hook.f., have been published validly by their authors. Engler (1921) considered B. longipetiolata to be conspecific with B. squamulosa , but as the result of the present study the former name is reinstated. The other three taxa: B. karperi J.C. Arends, B. pelargoniiflora J.J. de Wilde & J.C. Arends and B. rwandensis J.C. Arends, are new species described for the first time. All six species treated here are delimited from the other 24 species in the section Tetraphila by the combination of features mentioned in Chapter 1.Many features have been studied in preserved field collections, while quite a few data have been obtained from observations of living plants of B. elaeagnifolia , B. karperi , B. longipetiolata and B. squamulosa , both in the wild and in the greenhouse.From the review of previous taxonomic treatments in Chapter 2 it appears that, in the past, vegetative features have been overemphasized in the recognition and the delimitation of the taxa dealt with here. In Chapters 4 to 8 the results of detailed morphological, anatomical and karyological studies are discussed and compared with data from previous investigations in Begonia.Both B. elaeagnifolia and B. squamulosa comprise diploid as well as tetraploid populations. The 4x populations, which are interpreted to be interracial autopolyploids, are confined to the Crystal Mts in Gabon (Chapter 4).As to vegetative structures (Chapter 5), stems and leaves usually show few specific differences. For example, both the stem and the leaf blade in B. rwanden sis are thicker than those in the superficially very similar B. karperi. Both species have peltate leaves with an actinodromous venation, while the leaf venation in the other species is simple craspedodromous. The ratio of the length of the largest leaf blade and the length of the longest internode of a specimen is, in combination with reproductive features, often indicative for the identity of that particular specimen. The dimension and the shape of the leaf blades are hardly useful for the delimitation of the species recognized here.The leaf margins are provided with tiny teeth which are both 'violoid' and 'tylate'. The apices of the teeth often include cells with suberized walls. Because of further suberization of tissue of both the teeth and adjacent parts of the blade, there may be regions of wound cork on the margin of older leaves. Complex hydathodes, which are associated with tracheids from vein terminations, are present on the upper surface of the leaf margin, including the teeth. The petioles of B. longipetiolata and B. pelargonifflora are usually canaliculate, whereas those of the remaining species are more or less terete. The indumentum of all the plants studied consists of short-stalked dentate scales, which according to their shape form an intergrading morphological series.Several reproductive features are discussed in Chapter 6. As far as the number of flowers and the comparative length of the axes in the cymoid inflorescences are concerned, the species are usually distinct. The vascular anatomy of female flowers shows that the outer perianth segments are homologous with sepals and the inner ones with petals. The stamens have poricidal anthers. The 'front side' of the zygomorphic androecium in a lateral flower is always oriented towards the main inflorescence axis. The anther pores of the stamens on the front side of the androecium of B. longipetiolata have an adaxial orientation in relation to the main inflorescence axis, but in B. squamulosa their orientation is abaxial.Although specific trends in both the number of stamens and the number of styles can be discerned, stamen number as well as style number are not constant specific characters. The anthers in B. pelargoniiflora and B. rwandensis are longer than those in the other species studied. Except for B. longipetiolata, whose style surface below the style arms and the stigmatic band is more or less papillose, that surface in the other species is smooth. In B. squamulosa the smooth style surface carries some dentate scales.The ovaries investigated show both an axial and a parietal placentation in a single ovary. In B. longipetiolata, each septum with its two placental lobes is cross-shaped in transverse section, while, as far as could be studied, that in the other species is T- or arrow-shaped. The ovules in B. squamulosa have a pleurotropous position in the ovarial cavities, whereas those in the other species are epitropous. The number of styles does not always correspond with the number of septa and/or locules in the same gynoecium.The placentae in the ovaries in B. elaeagnifolia , B. longipetiolata and B. squamulosa are supplied by two almost independent vascular systems. The finding of two systems, viz.: an axial and a parietal system, is in line with the observation of Charpentier et al. (1989) that the ontogeny of the ovary in Begoniaceae is determined by both a single axial meristern in the ovary base and several parietal meristems on the ovary wall. The present study supports Reitsma's (1983) hypothesis that in section Tetraphila the septa are more or less 'placental' (Chapter 7).The investigation of pollen grains usually corroborates with Van den Berg's (1984) conclusions regarding the pollen morphology and pollen-type of the species studied here. However, this study indicates that there appears to be a transition between the B. squamulosa pollen-type and the B. komoensis pollen-type . Pollen grain size may be indicative for the ploidy level of a specimen of either B. elaeagnifolia or B. squamulosa (Chapter 8).In Chapter 9, it is discussed why the taxa recognized here are accorded the rank of species. With the exception of B. rwandensis, the species are sympatric and appear to have more or less similar ecological requirements. Various observations and crossing experiments indicate that the sympatric taxa, which may be found growing close together in the same locality, are reproductively isolated.Within their basic distribution several of the species have a localized topography that coincides with presumed African pleistocene rain forest refuges. It is inferred that gene flow among local populations is limited, while the possibility of gene flow among populations from geographically separated regions can be excluded. Finally, the phylogenetic relationships of the species are discussed.The taxonomic treatment in Chapter 10 includes a key to the species. They are described, typified and illustrated. Full synonymy, references to taxonomic literature and all the available material are cited. Distribution maps show where the species have been collected.