Under climate change, altered tropical cyclone regimes could cause long-lasting effects on coastal tropical forest structure, composition, and function. A pantropical meta-analysis of 48 case studies in 24 tropical forests affected by 20 cyclones indicated that total litterfall carbon (C) flux increased from ~1.2 ± 0.14 to 10.8 ± 1.44 g C/m2/day due to cyclones, and reached pre-disturbance levels within one-year post-disturbance. Predicting the large variation in cyclone response across pantropical sites, leveraging site-level ground and remote-sensing observations, is key to predictive modeling of tropical forest response to future cyclones.

Objectives / Hypothesis

To enable dynamic vegetation model predictions of post-cyclone forest biomass C fluxes, we combined ground and remote-sensing observations in a tropical forest affected by hurricanes Hugo and Maria. We compared ground and remote-sensing observations with simulated litterfall C flux and LAI. We expected that, by assigning different canopy damage and mortality rates to shade-tolerant and light-demanding plant functional types (PFTs), simulated litterfall and LAI data would match observations.


We compiled litterfall C flux and MODIS LAI 500-m data from Bisley forest (Puerto Rico) before and over 2 years after Hugo and Maria. We calculated hurricane-induced changes in litterfall C flux and LAI and compared with changes in litterfall C flux and LAI from simulations of the Functionally Assembled Terrestrial Ecosystem Simulator (FATES) coupled with the Energy Exascale Earth System Model (E3SM) land model (ELM-FATES). Following a 300-year spinup with ELM-FATES, we implemented hurricane disturbance with 100% defoliation, 20% structural biomass reduction, and varied mortality rates by referring to Hugo and Maria effects in Bisley.


By imposing 80% mortality of light-demanding and 50% of shade-tolerant PFTs to simulate the impact of a large hurricane on Bisley, simulated LAI decreased by 55.3%, while remotely-sensed LAI after Maria decreased by 59% relative to pre-hurricane levels. Ground observations suggested a 450-fold increase in total litterfall C flux post-Hugo, representing an instantaneous cyclone-caused input ~1.3 times the average annual litterfall of ~417 g C/m2/year. Changes in simulated litterfall C flux due to a large hurricane were fifteen times greater than ground observations following Maria, and 2.5 times smaller than observations following Hugo.

Implications / Conclusions

Ongoing work is focused on benchmarking ELM-FATES predictions against ground and remote-sensing data. To achieve reliable simulations of the effects of altered cyclone regimes on tropical forests, observational data and relationships provide critical benchmarks and inputs.


Dynamic vegetation model, Ecosystem function, Hurricane, Leaf area index, Litterfall.

Barbara Bomfim, Mingjie Shi, Dellena Bloom, Yanlei Feng, Michael Keller, Lara Kueppers

Presentation within symposium:

S-17 Linking field-oriented ecology and ecologists with land surface models and modelers

Leveraging multi-source forest data to predict cyclone responses in tropical forest