Recent syntheses of variation in woody plant traits have emphasized how hydraulic traits – those related to the acquisition, transport and retention of water across roots, stems and leaves – are coordinated along a limited set of dimensions or sequence of responses (Reich 2014, Bartlett et al. 2016). However, in many hydraulic trait-trait relationships, there is considerable residual variation, despite the fact that many bivariate relationships are statistically significant. In other instances, such as the relationship between root-stem-leaf vulnerability to embolism, data are so limited that testing the trait coordination hypothesis is not yet possible. The impacts on plant hydraulic function of competing hypotheses regarding trait coordination (or the lack thereof) and residual trait variation have not yet been comprehensively tested and thus remain unknown.

We addressed this knowledge gap with a parameter sensitivity analysis using a plant hydraulics model in which all parameters are biologically-interpretable and measurable plant hydraulic traits, as embedded within a size- and demographically-structured ecosystem model, the ‘Functionally Assembled Terrestrial Ecosystem Simulator’ (FATES). We focused on tropical forests, where co-existing species have been observed to possess large variability in their hydraulic traits. Assembling 10 distinct datasets of hydraulic traits of stomata, leaves, stems, and roots, we determined the best-fit theoretical distribution for each trait and quantified interspecific (between-species) trait-trait coordination in tropical forests as a rank correlation matrix. We imputed missing correlations with values based on competing hypotheses of trait coordination, such as coordinated shifts in embolism vulnerability from roots to shoots (the hydraulic fuse hypothesis). Based on the Fourier Amplitude Sensitivity Test and our correlation matrix, we generated thousands of parameter sets for an ensemble of hydraulics model simulations at a tropical forest site in central Amazonia. We explore the sensitivity of simulated leaf water potential and stem sap flux in the context of hypotheses of trait-trait coordination and their associated uncertainties.