Predictors of plant diversity in a hyperarid desert wadi ecosystem

The aim of this study was to identify trends in plant diversity, and examin e their predictive relationships with environmental variables (water stress , sail texture and grazing intensity) and Vegetation characteristics (total vegetation and individual species cover), along a 340 km transect through an area of hyper-arid desert in southern Egypt. The major part of the trans ect is located within the boundaries of the Wadi Allaqi Man and Biosphere ( MAB) Reserve. In total 55 plant species were recorded from the transect. A TWINSPAN classification of the data identified four of the seven main veget ation types, previously identified as present in the area by Springuel et a l. (1997). Sites dominated by Vegetation type I occurred exclusively in the lowest stretch of the Wadi Allaqi system, most heavily influenced by flood and seepage from Lake Nasser, and characterized by dense stands of Tamarix nilotica (Ehrenb.) Bunge and Glinus lotoides L. Vegetation type IIIb was r epresented at sites higher up the wadi system, characterized by sparser veg etation dominated by Acacia ehrenbergiana Hayne, Acacia tortilis subsp. rad diana (Savi) Brenan and Aerva javanica (Burm. f.) Juss.ex Schult. Vegetatio n type IIIa occurred in the middle region of Wadi Allaqi and was characteri zed by Salsola imbricata Forssk. and Cullen plicatum (Delile) C. H. Stirt. Vegetation type IV, the most species rich, was limited to upstream sites of Wadis Allaqi and Eiqat. Plant beta-diversity las species turnover between individual sites along the transect, assessed using Whittaker's beta(w) ind ex) showed a rising trend of plant species turnover from sites low in the w adi system to sites located in the upper part of the wadi. Plant alpha-dive rsity within these vegetation types (as species richness R: number of speci es present per 500 m(2) unit area) was linearly correlated with plant diver sity measured using Simpson's index (D). R was reasonably well predicted fr om three variables: plant cover, soil moisture and grazing intensity (follo wing a third order polynomial model). This model predicted R in the origina l data set with 52.9% success ( r = 0.73, p < 0.001). The study suggested t hat vegetation cover estimates, in association with simple estimates of int ensity of habitat stress (water availability) and disturbance (grazing inte nsity), can provide a rapid indication of the biodiversity support capacity of desert habitats. (C) 2000 Academic Press.