Despite the known links between climate and biogeochemical cycling of N in tropical forests, fundamental knowledge of N cycling is still far from complete. Our objective was to ascertain differences in the N cycle of two tropical dry forests under contrasting precipitation regime (1240 or 642 mm of mean annual rainfall). To do so, we examined a short-term metric of N cycling (N concentration) and a more integrated metric of N cycling (natural abundance 15 N) in plants and soils at both sites. At both the relatively wet and dry sites, N cycling associated with two non-N 2 -fixing species was compared to N cycling associated with two potential N 2 -fixing species; all four tree species considered were dominant at both sites. The 15 N abundance in plants was highest in the site with low rainfall, showing that N losses from the system may be large. By contrast, short-term N metrics did not vary with rainfall. Although there was a trend for leaf N concentration to be elevated in trees that have potential associations with N 2 -fixers, only 15 N in the forest floor was significantly greater under trees with high canopy N (N 2 -fixing species) than those with low canopy N (non-N 2 -fixing species). Within each site, the influence of N 2 -fixing species on N cycling increased with a reduction in rainfall. Overall, our results demonstrate the role of climate as a driver of N cycling in the region, such that the projected decrease in precipitation in this region may lead to larger N losses in these forests. This study also shows how changes in tree species with and without N 2 -fixing associations may impact N cycling in tropical dryland forests in the future.Abstract in Spanish is available with online material.

Aim: Spatial patterns in resource supply drive variability in vegetation structure and function, yet quantification of this variability for tropical dry forests (TDFs) remains rudimentary. Several climate-driven indices have been developed to classify and delineate TDFs globally, but there has not been a climo-edaphic synthesis of these indices to assess and delineate the extent of TDFs. A statistical climo-edaphic synthesis of these indices is therefore required. Location: Pantropical. Time period: Modern. Major taxa studied: Vascular plants. Methods: We assembled most known prior descriptions of TDFs into a single data layer and assessed statistically how the TDF biome, which we call tropical dry landscapes (TDLs) composed of forest and non-forest vegetation, varied with respect to the normalized difference vegetation index (NDVI) sensed by MODIS (250 m pixel resolution). We examined how the NDVI varied with respect to mean annual temperature (MAT) and rainfall (MAR), precipitation regime, evapotranspiration and the physical, chemical and biological properties of TDL soils.Results: Overall, the NDVI varied widely across TDLs, and we were able to identify five principal NDVI categories. A regression tree model captured 90% of NDVI variation across TDLs, with 14 climate and soil metrics as predictors. The model was then pruned to use only the three strongest metrics. These included the Lang aridity index, total evapotranspiration (ET) and MAT, which aligned with identified NDVI thresholds and accounted for 70% of the variation in NDVI. We found that across a global TDL distribution, ET was the strongest positive predictor and MAT the strongest negative predictor of the NDVI. Main conclusions:The remote sensing-based approach described here provides a comprehensive and quantitative biogeographical characterization of global TDL occurrence and the climatic and edaphic drivers of these landscapes.

Pantropical variation in soils of the tropical dry forest (TDF) biome is enormously high but has been poorly characterized. To quantify variation in the global distribution of TDF soil physical and chemical properties in relation to climate and geology, we produced a synthesis using 7500 points of data with gridded fields representing lithologic, edaphic, and climatic characteristics. Our analyses reveal that 75 TDF ecoregions across five biogeographic domains (Afrotropical, Australasian, Indo-Malayan, Neotropical, and Oceanian) varied strongly with respect to parent material: sediment (57%), metamorphic (22%), volcanic (13%), and plutonic (7%). TDF ecoregions support remarkably high variability in soil suborders (32), with the Neotropical and Oceanian realms being especially diverse. As a whole, TDF soils trend strongly toward low fertility with strong variation across biogeographic domains. Similarly, the exhibited soil properties marked heterogeneity across biogeographic domains, with soil depth varying by an order of magnitude and total organic C, N, and P pools varying threefold. Organic C and N pool sizes were negatively correlated with mean annual temperature (MAT) and positively correlated with mean annual precipitation (MAP). By contrast, the distribution of soil P pools was positively influenced by both MAT and MAP and likely by soil geochemistry, due to high variations in soil parent material across the biogeographic domains. The results summarized here raise important questions as to how climate and parent material control soil biogeochemical processes in TDFs.