An extensive suite of functional traits distinguishes Hawaiian wet and dry forests and enables prediction of species vital rates

The application of functional traits to predict and explain plant species’ distributions and vital rates has been a major direction in functional ecology for decades, yet numerous physiological traits have not yet been incorporated into the approach. Using commonly measured traits such as leaf mass per area (LMA) and wood density (WD), and additional traits related to water transport, gas exchange and resource economics, including leaf vein, stomatal and wilting traits, we tested hypotheses for Hawaiian wet montane and lowland dry forests (MWF and LDF, respectively): (1) Forests would differ in a wide range of traits as expected from contrasting adaptation; (2) trait values would be more convergent among dry than wet forest species due to the stronger environmental filtering; (3) traits would be intercorrelated within “modules” supporting given functions; (4) relative growth rate (RGR) and mortality rate (m) would correlate with a number of specific traits; with (5) stronger relationships when stratifying by tree size; and (6) RGR and m can be strongly explained from trait‐based models. The MWF species’ traits were associated with adaptation to high soil moisture and nutrient supply and greater shade tolerance, whereas the LDF species’ traits were associated with drought tolerance. Thus, on average, MWF species achieved higher maximum heights than LDF species and had leaves with larger epidermal cells, higher maximum stomatal conductance and CO2 assimilation rate, lower vein lengths per area, higher saturated water content and greater shrinkage when dry, lower dry matter content, higher phosphorus concentration, lower nitrogen to phosphorus ratio, high chlorophyll to nitrogen ratio, high carbon isotope discrimination, high stomatal conductance to nitrogen ratio, less negative turgor loss point and lower WD. Functional traits were more variable in the MWF than LDF, were correlated within modules, and predicted species’ RGR and m across forests, with stronger relationships when stratifying by tree size. Models based on multiple traits predicted vital rates across forests (R2 = 0.70–0.72; p