Maya Almaraz , Ecology and Evolutionary Biology, Brown University, Providence, RI
Harmony S. Lu , Ecology and Evolutionary Biology, Brown University, Providence, RI
Steven Goldsmith , Geography and Environment, Villanova University, Villanova, PA

Stephen Porder , Ecology and Evolutionary Biology, Brown University, Providence, RI

Background/Question/Methods
Many lowland tropical forests exhibit characteristics of nitrogen (N) richness – high rates of nitrification, mineralization and denitrification, high riverine losses of inorganic N, and enriched soil d15N. Nevertheless, tropical forests exhibit biogeochemical heterogeneity unparalleled at higher latitudes, and the drivers of this diversity remains poorly understood. We explored the relative importance of three potential drivers of N cycling in the Luquillo Experimental Forest of Puerto Rico: topography, parent material and forest community. We selected sixteen sub-watersheds in a full factorial combination of two parent materials (volcaniclastic and quartz diorite) and two forest communities (Colorado and Tabonuco). In each sub-watershed we sampled ridge, slope and valley positions. Within this framework, we measured soil C, N, and d15N. In the eight sites dominated by Tabonuco forests, we measured foliar d15N and N:P, KCl-extractable nitrate, ammonium, net nitrification and mineralization. In addition, we used acetylene block soil incubations to measure N2O and N2 production. We hypothesized that N cycling would differ across the landscape as a result of variation in soil texture (controlled by parent material) and topography (which influences soil saturation and oxygen availability), and that the different metrics of N availability would co-vary.
Results/Conclusions
Contrary to our hypothesis, we found that the interpretation of landscape scale variation in N cycling was dependent on the metric used. Soil C:N was higher, and d15N was lower, in Colorado compared with Tabonuco forests (n=16, p<0.002). Since forest type and elevation are correlated across this landscape, we tested this relationship in eight mid-elevation sites where that correlation breaks down, and found these community-driven differences persisted (n=8, p<0.006). Within Tabonuco forests, net mineralization (p=0.03) and soil d15N (p=0.0003) were higher on clay-rich volcaniclastic-derived soils than on sandier quartz diorite-derived soils, regardless of topographic position. In contrast, N gas emissions (N2O+N2) were marginally higher on quartz diorite-derived soils (p=0.085), and N2 emissions were marginally higher in valleys (p=0.079), but N2O emissions did not vary across the two parent materials (p>0.1). Topography, not parent material, best explained the variance in foliar N:P of Tabonuco (Dacryodes excelsa) trees, which was lowest in valleys (p=0.014). We conclude that N cycling varies across this landscape, but that different processes are influenced by different landscape-scale parameters. This highlights the challenge of scaling plot-scale metrics of N cycling up to a landscape-level description of ecosystem N status.