Recent work has pointed toward the tremendous success of legumes in recently disturbed, water-limited tropical forests. This success has been tied to the ability of many legumes to fix atmospheric dinitrogen. One way to test causal linkages between legume success and nitrogen fixation is through mechanistic dynamic vegetation models. However, this approach is challenging because such models have rarely included legumes as a distinct plant functional type (PFT). Here, we incorporated legume PFTs into the dynamic vegetation model ED2 and applied the model to the problem of secondary succession in Neotropical dry forests. Legume PFTs were different from non-legume PFTs only in that they are able to fix nitrogen (at a carbon cost). We addressed three questions. (1) What are the model biases along successional and soil fertility gradients, and do legume PFTs reduce model biases? (2) What factors control the response of legumes and non-legumes to nutrient fertilization treatments? (3) How does legume presence affect regional carbon and nitrogen budgets in wet versus dry sites? Model results were compared to both chronosequence observations and a nutrient fertilization experiment in Guanacaste, Costa Rica. Regional-scale simulations were done to better understand how the ecological role of legumes interacts with rainfall.
Results/Conclusions
Along successional and soil fertility gradients, our control simulations without legume PFTs underestimated plant biomass and overestimated litter carbon relative to observations. Both biases were reduced by the addition of legume PFTs, but not completely eliminated. In simulations of a nutrient fertilization experiment, the simulated biomass response of legumes to phosphorus addition was much larger than the response of non-legumes. This result is in accord with an actual, ongoing nutrient fertilization experiment in Guanacaste. In the model, the result occurs because dinitrogen fixation scales with phosphorus availability. Finally, across a rainfall gradient in Central America, we found that the effects of nitrogen fixation on ecosystem carbon and nitrogen storage were smallest at the driest sites because those sites were limited more by water than by nutrients. Collectively, our results highlight some of the challenges in incorporating a legume PFT into dynamic vegetation models. Incorporation of a fixation trait, by itself, is not guaranteed to translate into dramatic changes in tropical ecosystem carbon and nitrogen economies; rather, effects may be contingent on factors related to phosphorus, rainfall, or other processes. Further work is needed to determine other legume traits that can drive legume success in recovering tropical dry forests.