Questions: Premontane tropical forests harbour exceptionally high plant species diversity; understanding which factors influence their species composition is critical to conserving them, and to predicting how global environmental change will affect them. We asked: (1) how do aand b-diversity vary at the landscape scale; (2) how important is environmental filtering in structuring these communities; and (3) which soil and climate variables account for the most compositional variation?Location: Old-growth premontane forest, Fortuna Forest Reserve, western Panama.Methods: All trees ≥5-cm DBH were censused in 12 1-ha plots up to 13 km apart. For each plot, we measured soil properties (0-10 cm depth) at 13 locations, and estimated or measured monthly rainfall. To evaluate how the environmental and spatial variables are associated with community composition, we used ordination and Mantel tests.Results: Diversity varied nearly three-fold among plots (68-184 speciesÁha À1 ).b-Diversity was also high, with only one of 364 species present in all plots. Turnover reflected distinct forest community types that have developed on different parent materials: forests on rhyolite had an abundance of either ectomycorrhizal-associated trees or canopy palms, while forests on the other rock types (andesite, dacite and basalt) were dominated by trees that form arbuscular mycorrhizal associations. While NMDS ordination showed that species turnover was significantly correlated with rainfall seasonality, and also co-varied with geographic distance. Nonetheless, large compositional differences were apparent among sites <2 km apart with similar rainfall but differing soils. Partial Mantel tests controlling for geographic distance highlighted the relationship between total phosphorus and species composition.Conclusions: Soil nutrient availability and rainfall seasonality in premontane forests at Fortuna are associated with striking variation in the taxonomic and functional composition of nearby tree communities, and with plot differences in species richness comparable in magnitude to those reported over >1000 m a.s.l. in previous studies. Accounting for how local edaphic conditions structure premontane and montane tropical forests will be critical to predicting how tree communities will respond to climate change.

Fertilization experiments in tropical forests have shown that litterfall increases in response to the addition of one or more soil nutrients. However, the relationship between soil nutrient availability and litterfall is poorly defined along natural soil fertility gradients, especially in tropical montane forests. Here, we measured litterfall for two years in five lower montane 1-ha plots spanning a soil fertility and precipitation gradient in lower montane forest at Fortuna, Panama. Litterfall was also measured in a concurrent nitrogen fertilization experiment at one site. Repeated-measures ANOVA was used to test for site (or treatment), year, and season effects on vegetative, reproductive and total litterfall. We predicted that total litterfall, and the ratio of reproductive to leaf litterfall, would increase with nutrient availability along the fertility gradient, and in response to nitrogen addition. We found that total annual litterfall varied substantially among 1-ha plots (4.78 Mg/ha/yr to 7.96 Mg/ha/yr), and all but the most aseasonal plot showed significant seasonality in litterfall. However, litterfall accumulation did not track soil nutrient availability; instead forest growing on relatively infertile soil, but dominated by an ectomycorrhizal tree species, had the highest total litterfall accumulation. In the fertilization plots, significantly more total litter fell in nitrogen addition relative to control plots, but this increase in response to nitrogen (13%) was small compared to variation observed among 1-ha plots. These results suggest that while litterfall at Fortuna is nutrient-limited, compositional and functional turnover along the fertility gradient obscure any direct relationship between soil resource availability and canopy productivity.Abstract in Spanish is available in the online version of this article.