LANDSCAPE AND MICROSITE INFLUENCES ON SHRUB RECRUITMENT IN A DISTURBED SEMIARID QUERCUS-JUNIPERUS WOODLAND

Seedling demography of the leguminous shrubs Leucaena pulverulenta and Acacia berlandieri was monitored to determine the functional role of microsites in the restoration of disturbed semiarid woodlands. Seeds w ere sown on replicated landscapes during 1989 and again in 1990 into f our microsites after brush clearing: 1) Rocky, with shallow soils (<10 cm) and low vegetative cover (much less than 50%), 2) Herbaceous, wit h shallow soils (10 to 20 cm) and nearly 100% cover of perennial short -grasses, 3) Tree, with deep soils (20-30 cm) and 5-10 cm of duff unde r Quercus virginiana/Juniperus ashei mottes, and 4) Cleared, resulting from mechanical removal of Quercus/Juniperus mottes, with deep soils (20-30 cm) and 2-10 cm of duff. Within a landscape, recruitment was gr eatest on Cleared and Tree (>80%), intermediate on Rocky (65%), and lo west on Herbaceous (36%) microsites. Desiccation appeared to be the pr imary cause of first-season mortality. Differences between thr microsi tes in seedling height and number of leaves were significant the first planting, bur nor the secund. Overwinter mortality of Leucaena and Ac acia seedlings was lowest on Cleared microsites (55 and 7%, respective ly) and greatest on microsites dominated by Herbaceous vegetation (100 and 87%, respectively). Survival of established seedlings during the secund growing season exceeded 80% for both species on all microsites. A. belandieri had the highest establishment potential and the greates t ecological amplitude, with recruitment after two years being compara ble on Cleared, Rocky and Tree microsites (72-78 seedlings). L. pulver ulenta showed marked preferences for Cleared microsites (47 seedlings established) relative to Rocky (13 seedlings), Tree (12 seedlings) and Herbaceous microsites (0 seedlings). Spatial variability in emergence between replicated landscapes (range 35 to 78%) was greater than the variability between microsites within a landscape (48 to 62%). Differe nces in seed disappearance associated with surface runoff from high in tensity storms and losses to granivory on Tree microsites were the pri mary causes of differences in recruitment between landscapes. Inter-la ndscape variability in granivory and precipitation runoff therefore ov ershadowed the effects of within-landscape seed placement among first year seedlings. Generalizations of seedling ''safe site'' characterist ics based solely on short-term (one year), pseudoreplicated, within-la ndscape studies may therefore be misleading. Failure to account for va riability in important processes at greater spatial scales may signifi cantly influence the robustness of experimental microsite study result s derived from small-scale research.