Informing whole-tree performance under drying conditions using an easily measurable leaf-level trait in a tropical rainforest

Isabelle Maréchaux, Damien Bonal, Megan K. Bartlett, Benoît Burban, Sabrina Coste, Elodie A. Courtois, Maguy Dulormne, Jean-Yves Goret, Eléonore Mira, Ariane Mirabel, Lawren Sack, Clément Stahl, Jérôme Chave
Variation in water availability is a key driver of forest ecosystem function and tree species distributions. However, our knowledge of tree responses to changes in water supply and the diversity of these responses across species is still incomplete. This hampers our ability to make informed predictions, especially given the forecasted increase in rainfall variability and drought frequency under climate change. This challenge is amplified in tropical forests, which shelter globally important stores of biodiversity and carbon, play a critical role in rainfall recycling through evapotranspiration, and appear to be vulnerable to drought.

The high taxonomic and functional diversity and the typical absence of species dominance in tropical forest communities call for a trait-based approach. However, the quest for relevant and reliably measurable functional traits that quantify plant performance under drying conditions still remains a major challenge of plant functional ecology as plant responses to decreasing water availability result from a complex interplay of mechanisms operating across scales within the plant.

In this study, we tested the hypothesis that the leaf water potential at turgor loss or wilting point, which determines the tolerance of leaves to drought stress, informs tree performance under drying conditions. To do so, we monitored the rate of sap flow, which is an integrated measurement of water use and transpiration at the whole-plant scale, on trees in an Amazonian rainforest during a harsh dry season.

Our empirical data confirmed this hypothesis, hence providing an explicit link between a trait, measurable at the leaf level, and whole-plant performance under adverse drying conditions. Our results should help vegetation modellers improve their models and therefore gain confidence in the quality of their predictions.

Read the paper here.