RESUMEN
Successful persistence of dry forests depends on tree regeneration, which depends on a balance of complex biotic interactions. In particular, the relative importance and interactive effects of shrubs and herbivores on tree regeneration are unclear. In a manipulative study, we investigated if thornless shrubs have a direct net effect, an indirect positive effect mediated by livestock, and/or an indirect negative effect mediated by small vertebrates on tree regeneration of two key species of Chaco forest (Argentina). In a spatial association study, we also explored the existence of net positive interactions from thorny and thornless shrubs. The number of Schinopsis lorentzii seedlings was highest under artificial shade with native herbivores and livestock excluded. Even excluding livestock, no seedlings were found with natural conditions (native herbivores present with natural shade or direct sunlight) at the end of the experiment. Surprisingly, seedling recruitment was not enhanced under thornless shrubs, because there was a complementary positive effect of shade and interference. Moreover, thornless shrubs had neither positive nor negative effects on regeneration of S. lorentzii. Regeneration of Aspidosperma quebracho-blanco was minimal in all treatments. In agreement with the experiment, spatial distributions of saplings of both tree species were independent of thornless shrubs, but positively associated with thorny shrubs. Our results suggest that in general thornless shrubs may have a negligible effect and thorny shrubs a net positive effect on tree regeneration in dry forests. These findings provide a conceptual framework for testing the impact of biotic interactions on seedling recruitment in other dry forests.