The morphology, anatomy and pollination biology of Mystacidium Lindl., a small, epiphytic genus of orchids, was investigated within a phylogenetic context. Morphological and anatomical studies were carried out in order to obtain characters for a cladistic analysis of the genus using Cyrtorchis arcuata (Lindl.) Schltr. as an outgroup. The phylogenetic analysis indicated that the genus may not be monophyletic. Two species of the closely related genus Diaphananthe Schltr., D. caffra (H.Bol.) Linder and D. millarii (H.Bol.) Linder, appear to be nested within Mystacidium. Mystacidium species grow in habitats varying from mistbelt forest to dry savanna. Analysis of stable isotope composition (Ȣ¹³C values) of leaves and roots showed that all Mystacidium species, as well as D. caffra and the outgroup C. arcuata, employ CAM photosynthesis. The Ȣ¹³C values were significantly negatively correlated with mean annual rainfall at the collection sites. Breeding system experiments revealed that Mystacidium is dependent on pollinators for fruit set, and that self-pollination results in substantially reduced seed set due to either inbreeding depression or partial self-incompatibility. Field observations revealed that M. venosum Harv. ex Rolfe and M. capense (L.f.) Schltr. are hawkmoth-pollinated, and that M. gracile Harv. and M. pusillum Harv. are pollinated by settling moths. The spurs of the flowers contain dilute, sucrose-dominant nectar. Mystacidium venosum and M. capense showed evidence of nectar reabsorption. Nocturnal emission of scent occurred in all species except M. aliceae H. Bolus and M. brayboniae Summerh., which are unscented, and was composed largely of a combination of monoterpenes and benzoids. Despite substantial variation in spur length (1 - 4.7 cm) among species, no evidence for directional selection on spur length was found in M. venosum, M. capense or M. gracile. Hand pollinations significantly increased fruit set in M. capense in two consecutive seasons at different sites, indicating pollen limitation. Although pollen removal was greater than pollen receipt in M. venosum, M. capense and M. gracile, suggesting transport loss or insufficient visitation, a remarkably high percentage of removed pqllen reached stigmas (35 - 50%). Experiments on M. venosum revealed that flower longevity is reduced by pollination, and that pollinia removed from flowers remained viable for up to 20 days under field conditions. The phylogeny indicated that long-spurred, hawkmoth-pollinated species are basal within the genus, and that shorter-spurred species pollinated by noctuid moths are derived. / Thesis (M.Sc.)-University of Natal, Pietermaritzburg, 2002.