We investigated the effects of host tree on epiphyte diversity, controlling for microclimate. We measured the light profiles of the lower trunks of 20 individuals, each from three host groups (tree ferns, dicots, palms) occupying the understorey in a tropical montane forest in Panama. The per cent cover and species richness of vascular and non-vascular epiphytes were surveyed on the lower trunks of each understorey host. Light varied considerably between trees (5–21% total transmitted light) but mean light level did not vary between types of host. Light was not significant as a covariate with host in any model. Tree ferns had higher covers than dicots and palms of filmy ferns (15%, 0.02% and 0.2%), other ferns (7%, 0% and 0.5%) and other vascular epiphytes (16%, 3% and 3.4%), and greater species richness of vascular epiphytes (filmy ferns: 3, 0.4 and 0.5; other ferns: 2, 0.2 and 0; other vascular: 7, 2 and 2). Dicots had a higher cover of liverworts (53%) than palms (18%) and tree ferns (27%). Palms and tree ferns were the compositional extremes. We conclude that the differences in species composition and cover between the three host groups relate better to physical differences between hosts than differences in light climate.

In this study, we determined the effects of plant richness, total plant abundance and abundance of plants with high galling diversity (super-hosts) on the local and regional galling richness in 15 areas of Neotropical savanna in different parts of Brazil. We established in each area 10 plots of 10 × 10 m, where we sampled a total of 131 plant species and 112 galling insect species. We considered all areas at the regional level and plots within each area at the local level. We found that at the regional level a combination of plant richness and plant abundance is the best model to explain the gall-inducing richness. We also found that at the local scale the most important factors are plant abundance and super-host abundance, but the plant richness is irrelevant. These findings confirm recent results that plant richness is an important predictor of galling species richness at the regional level, whereas the vegetation composition and structure act at the local level. Furthermore, this study contributes to the knowledge of the ecology of galling by showing how local factors influence the richness of this insect guild.