Introduction / Background / Justification: Tropical forests cycle more carbon, water and energy than any other biome, such that any perturbation in its function has repercussions on the global climate. Droughts and extreme wetness are projected to increase in tropical forests depending on region, but how those changes will affect tropical forest composition and function remains uncertain. Next-generation vegetation demography models (VDMs), a component of land surface models, are based-on performance-trait-environment interactions and promise to improve our predictive power, but these interactions are challenging to quantify, especially given limited data for trees' belowground traits and water environments.


Objective(s)/Hypothesis(es): Evolution and community assembly processes dictate that the extent of trees’ response to droughts, or extreme wetness, depends on the perturbation of species' historical hydrological environment and trait-adaptations. Here, I showcase how hydrology, demography and traits datasets can be combined to improve our understanding of plants' drought-response strategies to advance land surface models.


Methods: For a moist tropical forest in Barro Colorado Island, Panama, we parameterised a next-generation vegetation demography model coupled with a land surface model, ELM-FATES, by leveraging local soil hydraulic properties and by maximizing the fit between ensemble simulations and observations for multiple hydrological states and fluxes simultaneously. Droughts were identified as anomalies in soil water potential dynamics over the whole soil column spanning three decades of tree demographic censuses. We developed a novel statistical model to estimate species-specific effective rooting depths based on growth records, concurrent soil water-potentials and leaf hydraulic vulnerability curves. Model estimates were calibrated against isotopes-based water-sourcing depths. We also found and leveraged that soft functional traits such as LMA and wood density were effective predictors of leaf vulnerability curves.


Results: This study revealed that drought response is a function of drought-exposure relative to trait-adaptations at the whole-plant level: deep rooted species were associated with more vulnerable stem hydraulic traits, yet as a result of reduced hydrological drought-exposure they had lower mortality rates than shallow rooted species through several El-Nino droughts.


Implications/Conclusions: These finding imply that if VDMs are only driven by above-ground hydraulic traits they would kill the wrong trees under droughts by implementing their greater risk of embolism; VDMs must also take rooting depths into account to ascertain if embolism risk is realized via drought-exposure. I will highlight the data required to rapidly advance the capacity of VDMs to predict forest response to droughts across the tropics and beyond.