COS 37-6 - A mechanistic model for tree-liana interactions and canopy dynamics

Background/Question/Methods

Lianas (woody vines) compete with trees for light and other resources. Some observations and liana-removal experiments have shown that liana-colonized trees can have significantly lower growth rates and higher mortality rates than uncolonized trees, leading to reduced forest carbon accumulation. Despite the importance of lianas for forest ecology and carbon budgets, very few mechanistic forest dynamics models have explicitly included lianas. To help fill this gap, we first present a model that contains a unique and explicit representation of lianas. Then, we use the model to answer the following: (i) Which tree and liana traits exert the strongest influence on simulated tree-liana competition? (ii) How does the inclusion of lianas affect simulations of tropical forest dynamics? Our starting point is the TROLL model, an individual-based model of forest dynamics. TROLL can include an arbitrary number of tree species, and species are defined by a modest number of traits. The model is spatially explicit on the scale of one cubic meter; thus, canopies are explicitly resolved. We have added lianas to TROLL and have carried out simulations corresponding to a field site in Guanacaste, Costa Rica.

Results/Conclusions

TROLL now includes lianas. A liana organism can have zero, one or multiple aboveground stems. Each stem may or may not be colonizing a tree canopy. In our sensitivity analysis, we found strong sensitivity to liana leaf production schemes. Simulated lianas were rarely viable unless they preferentially produced leaf area at the top of their host tree canopy. Simulations were also very sensitive to the maximum amount of leaf area per unit volume. Other traits determined whether colonization would lead to tree mortality. Once colonized, early successional tree species were more likely to die than late successional tree species. In addition, colonizing lianas having high leaf mass per unit area (LMA) were more likely to lead to tree death the lianas with low LMA. Simulations with and without lianas differed significantly. TROLL without lianas simulated about 20% more biomass than TROLL with lianas. In addition, simulations without lianas had fewer stems, higher mean wood biomass, and less diversity than simulations with lianas. We advocate for increased measurements of liana traits, which will enable better parameterization of models. Improving models in this way is essential given the likely large impacts of lianas on tropical forest structure, functioning, and composition.