COLONIZATION BY ANNUAL PLANTS OF AN EXPERIMENTALLY ALTERED DESERT LANDSCAPE - SOURCE-SINK RELATIONSHIPS

1. We studied how annual plant species colonize a set of patches (natu ral and human-made) in a shrubland landscape in the Negev desert. We a sked what patch properties and species' attributes affect colonization during 4 years after experimental formation of eight sets of pits and mounds. 2. The 'sink function' of patches was measured as the numbers of colonizing species, the proportion of patches colonized during eac h year, and the abundance of the colonizers. The effects of species' d ispersal I-node and propagule size were examined. 3 In the first growi ng season, pits and mounds were colonized by large numbers of species. The total number of colonizing species was similar for the two patch types, but incidence was higher in pits than in mounds. 4 The early co lonizers were mainly wind-dispersed and showed both high abundance and incidence. Dispersal mode did not affect colonization in the second y ear and later, nor did propagule size throughout the experiment. 5 Mos t early colonizers became resident in the patches, but pits and mounds showed some further colonization during the second year. In the third and fourth year both the numbers of colonizing species and their inci dence decreased. 6 Pits were mainly colonized from outside the experim ental units to which they belonged, whereas mounds received species mo stly from the adjacent pits within the same units. 7 From the second y ear on, patches of undisturbed crusted soil were also colonized, mainl y from the adjacent pits and mounds. The number of colonizing species in undisturbed patches increased during the last, high-rainfall, year both per patch and for the whole set of patches. 8 We conclude that th e sink function of patches for colonization increases with removal of vegetation, soil crust and seed bank, with patch capacity to capture r esource and plant propagules, and with amount of rainfall. Sink functi on decreased with time due to decreasing availability of new species, but increased with the presence of stronger sink patches in the vicini ty. 9 We suggest that the concept of sink function, in terms of the nu mbers, incidence and abundance of species colonizing sets of patches, can be a powerful tool for linking community and landscape processes.