We assessed soil seed bank composition and size over several microhabitats of two habitats of the central Monte Desert of Argentina (open Prosopis woodland and Larrea shrubland) to analyse differences among them. Seed densities were similar to those already reported for other deserts, but we found consistent differences in seed composition among microhabitats. Whereas grass seeds (e.g., Aristida, Pappophorum, Neobouteloua, Trichloris, Digitaria) prevailed in natural depressions of open areas, forb seeds (e.g., Phacelia, Lappula, Descurainia, Plantago, Chenopodium) were more abundant under trees. The comparison of seed production during primary dispersal (i.e., seed rain) with seed density on the ground at the end of dispersal indicated that most forb seeds entered the habitat through the micro habitats located beneath the canopy of trees and tall shrubs, and remained there after redistribution. Most grass seeds, by contrast, entered it through bare-soil and under-grass microhabitats, and reached more even distributions after secondary dispersal, especially because of dramatic losses in bare soil. Patterns of plant recruitment and seed dynamics in specific microhabitats were better understood when differences of soil seed bank composition, but not of total seed density, were taken into account.Key words: Monte Desert, seed dispersal, seed predation, seed production, seeds.

No abstract

Summary1 Patterns of seed germination of grass and forb species were studied in open Prosopis woodland of the central Monte desert (Argentina) during several years, to test the hypotheses that (i) seed germination is positively aected by both rainfall and protection aorded by vegetation cover (a facilitative eect), (ii) the number of surviving plants is positively in¯uenced by rainfall but negatively aected by established vegetation (a competitive eect), and (iii) seed loss from soil banks owing to germination is lower than that caused by granivorous animals. 2 Forb species germinated during restricted periods, either in early autumn or in spring. Grasses, however, germinated throughout the growing season, but because seedlings could not be identi®ed to species level, it was impossible to discern whether dierent species germinated in particular seasons, or if all grasses germinated in all seasons. Grass and forb germination were generally of similar magnitude, but grass germination increased by an order of magnitude during a summer of unusually abundant rainfall related to an El NinÄ o Southern Oscillation (ENSO) event.3 Overall, the spatial distribution of neither germinating seeds nor surviving plants could be explained by interactions with established vegetation (facilitation and competition eects, respectively). An alternative explanation may be provided by the distribution of forb and grass seeds in the soil. 4 Seed loss owing to germination was low in both dry and rainy years. For forbs, such loss totalled < 1% of soil-seed reserves, and no forb species suered losses > 4%. Total grass-seed loss to germination was usually < 0.5%, and the 5% reached in 1997±98 corresponded to an interruption of a prolonged drought by unusually abundant rainfall associated with a reduced seed bank. 5 Grass-seed loss caused by germination was one to two orders of magnitude lower than that reported due to autumn-winter granivory in the central Monte desert.

Harvester ants usually go through temporal fluctuations in environmental seed abundance and composition which could influence their behaviour and ecology. The aim of this study was to evaluate how these fluctuations influence the diet of Pogonomyrmex rastratus, P. pronotalis and P. inermis (Hymenoptera, Formicidae) in the central Monte desert during three consecutive growing seasons. Although seeds were the main item in the diet, these ants turned more generalist when seed abundance of the most consumed species (grasses Aristida spp., Trichloris crinita, Pappophorum spp., Digitaria californica and Stipa ichu) was low. Accordingly, diversity of items in the diet decreased with seed abundance in a logarithmical fashion, showing higher foraging efficiency for seeds at higher seed abundance. Seed diversity, however, was not related to seed abundance as ants always included several species in their diet, with alternating prevalence. The proportion of the most consumed species increased logarithmically in the diet of P. rastratus and P. pronotalis along with their abundance in the environment probably as a consequence of diet switching (from forb and shrub seeds to grass seeds) and by an increase in foraging efficiency at higher seed densities. In contrast, foraging activity of P. inermis was very low at low seed abundance and its diet included only the five grasses. Among the most consumed species, proportion in the diet was not associated with relative abundance in the environment. Aristida spp., Pappophorum spp. and D. californica were overall highly selected. However, the flexibility in the diet of P. pronotalis and P. rastratus and the low foraging activity of P. inermis during periods of low resource abundance could attenuate potential top-down effects in the central Monte desert. This study shows that bottom-up effects are important in ant-seed interactions and should be considered when predicting and evaluating ants' effects on seed resources. Seed abundanceSeed abundance of the grasses Aristida spp., T. crinita, Pappophorum spp., D. californica and S. ichu, the species SEED ABUNDANCE FLUCTUATIONS AND ANT DIET 909

No abstract