Flora of the Pro-Namib Desert Swakop River catchment, Namibia: Community classification and implications for desert vegetation sampling

Plant assemblages were studied in the Pro-Namib Desert Swakop River catchment in Namibia (southern Africa) in order to describe the communities present. Two surveys were conducted: a stratified survey of desert and dry riverbed vegetation and a grid-based survey of desert vegetation alone. In the first survey, 'very high' growth forms (plants > 150 cm height) were recorded using eight 50 x 50 m quadrats in each of four habitats. In the second survey, all plants except grasses were split into high (plants > 50 cm) and low (plants 50 cm) growth forms and each recorded using 25 quadrats on five 2.4 km transects (one per 600 m of transect) placed in parallel, 2 km apart (with minimum quadrat size determined by the nested quadrats method). Data from both surveys were analysed using standard classification (TWINSPAN) and ordination (DECORANA) techniques. These analyses were conducted for species and quadrats, both (1) with and without outliers (both cases), and (2) with and without downweighting of rare species (species only). However, the ecological interpretation of the results remained consistent throughout. Results from the stratified survey revealed distinct plant communities; (1) riparian woodland characterized by Prosopis glandulosa and Salvadora persica, and (2) desert scrub characterized by Acacia reficiens, Commiphora glaucescens and C. virgata. Sub-communities in desert scrub were also recognized; these were plain and hill assemblages, distinguished by the presence or absence of Phaeoptilum spinosum, respectively. These communities match geomorphological features of the desert landscape and may reflect differences in the availability of moisture. However, these communities were largely undetectable in the analyses of the grid-based survey dataset, which suggests that the scale of that survey was inadequate for the purpose of classification, despite the attainment of plateaus on the species-area curves for quadrats. These results therefore draw attention to the problems of rarity and scale in the study of desert plant communities. (C) 1997 Academic Press Limited.