Seed removal and fate in two selectively logged lowland forests with constrasting protection levels

We evaluated seed removal by terrestrial mammals and the fate of removed, t hreaded seeds (as a measure of dispersal) in two neighboring tropical rain forest sites in the Caribbean lowlands of Costa Rica. Both sites have been selectively logged but differ in their degree of protection from human intr usion and habitat connectivity: La Selva is protected from hunting and conn ected to a national park, whereas Tirimbina remains unprotected and is not connected to a park. The two study sites are similar in logging intensity, elevation, and canopy tree structure and composition. We predicted that rat es of seed removal and seed dispersal would be higher at the connected and protected site. For seven tree species (six of which are timber species), w e determined that patterns of seed removal under an exclusion experiment (s emipermeable cages vs. uncaged) varied both within species across sites and within sites across species, suggesting site differences in abundance, deg ree of animal activity, or presence of particular mammal seed consumers. Ra tes of seed removal and dispersal were largely species-specific. Most of th e study species showed neither site nor treatment effect, whereas others ha d disproportionately higher removal rates at La Selva. For all species comb ined (covering a 6-month period due to phenological differences among the s tudy species; n = 920 seeds per site), twice as many removed seeds were dis persed at La Selva (5.3%) than at Tirimbina (2.2%). Only one timber species , Pentaclethra macroloba, did not appear sensitive to site differences in r ates of seed removal, most likely because its seeds are toxic to animals. T wo timber species, Carapa nicaraguensis and Lecythis ampla, whose seeds are consumed by large scatterhoarding rodents, had 2.5% and 13%, respectively, of their removed seeds dispersed at La Selva, whereas no seeds were disper sed at Tirimbina. In northeastern Costa Rica, where forest cover is fragmen ted and selective logging is currently underway, the biological sustainabil ity of timber species dispersed by terrestrial mammals may be more likely i n sites protected from hunting and/or adjacent to protected areas than in f orested patches subjected to hunting.