Tuesday, August 7, 2007

PS 38-193: Seed predation as a biotic filter for tree recruitment in successional shrublands in Chiloé Island, Chile

Marcela A. Bustamante-Sánchez1, Ariel A. Farias2, and Juan J. Armesto1. (1) CASEB, P. Universidad Católica de Chile. IEB, Universidad de Chile, (2) CASEB, P. Universidad Católica de Chile

Because recruitment in plant populations is a multiphase sequential process, an integrated view of the relationships between stages during the regeneration processes is needed. Biotic and abiotic factors have been often found to limit the number and spatial distribution of recruits within stands. However, few studies have experimentally assessed the direct and indirect effects of both established vegetation structure and rodent activity on recruitment. We used structural equation modeling to investigate the direct and indirect (i.e. mediated by rodent activity) effects of microhabitat structure (e.g. cover and volume of vegetation strata) on seed predation in successional shrublands of the rural landscape of Chiloé Island. We collected information on seed removal, micro-habitat use by rodents and vegetation structure, using a sampling grid (90x50 m). We established 60 sampling points (10 m apart) set along 6x10 parallel transects in three types of shrublands, differing in cover, vegetation height and composition. To estimate seed removal, Petri dishes were placed in each point, each containing 10 wheat seeds. Habitat use by rodents was estimated from capture frequency during five nights, in 180 Sherman traps placed by the sampling points. Overall, we found no direct effect of rodent capture frequency on seed predation rate.While seed predation rates responded positively to vegetation cover < 1 m of heigh (small shrubs and ferns), rodent capture frequency was negatively associated with the presence of green trees and positively to logs in the shrublands. The variables that affected seed predation also determined differentiation among shrublands: that with a higher proportion of microhabitats dominated by vegetation < 1 m of height showed higher seed predation relative to the other two shrublands (40% vs 12-13%). Consequently, despite the absence of a direct rodent effect, microhabitat distribution among the different types of shrublands could nonetheless affect the recruitment through seed predation.