The more traits the merrier: using an extensive suite of functional traits to characterize forests and predict tree growth and mortality

Camila D. Medeiros, Christine Scoffoni, Grace John, Megan Bartlett, Faith Inman-Narahari, Rebecca Ostertag, Susan Cordell, Christian Giardina, Lawren Sack

Traits predict tree species forest distributions and growth and survival rates across Hawaiian montane wet forest (MWF) and lowland dry forest (LDF). The top and bottom rows compare the MWF and LDF respectively, with forest images in left-hand column, and micrographs of impressions of the abaxial leaf surfaces and of cleared and stained leaves in the middle and right columns respectively (sides of squares = 0.31 mm and 1.25 mm respectively) for representative species Coprosma rhynchocarpa of the MWF and Psydrax odorata of the LDF. On average across species, stomatal density is higher for species of the MWF, and minor vein density is higher for species of the LDF, and these traits predict vital rates across all species. Photo credit: left, top and bottom: CTFS; center and right, top and bottom: Jessica Smith

For decades, ecologists have focused on relatively easily measurable functional traits in an attempt to predict species’ habitat distributions and their vital rates, such as relative growth rate (RGR) and mortality (m). Most studies focused on few functional traits, and achieved limited ability to resolve trait differences between forests or to predict vital rates. We hypothesized that the traits typically used most in previous research, while important, are often not sufficient to capture important axes of variation in plant function that explain higher-order ecological processes. Therefore, we measured an extensive suite of 45 functional traits, including stomatal, venation and hydraulic traits to determine whether a broader set of traits would provide valuable predictive power.

We tested key hypotheses derived from first principles in trait physiology and ecology in forest plots in montane wet forest (MWF) and lowland dry forest (LDF) of Hawai‘i in which all trees are mapped, tagged and censused repeatedly for growth and survival. We tested whether the species of the two forests differed broadly in the suite of traits based on their contrasting adaptation, how traits were inter-related, and if this suite of traits could statistically predict tree growth and mortality across forests.

Confirming our hypotheses, MWF species showed adaptation to high soil moisture and nutrient supply and greater shade tolerance whereas the LDF species showed adaptation to drought tolerance. Traits also tended to be inter-correlated into clusters. Many traits were correlated with species’ vital rates across forests, with stronger relationships when accounting for tree size variation within species. Models based on multiple traits provided unprecedented power to predict tree RGR and m across forests.

The use of an extensive suite of functional traits provides strong power to resolve ecological patterns and predict differences across species in growth and mortality. We conclude that a broad range of functional traits contribute fundamentally to species’ forest distributions and vital rates. We propose as an exceptionally promising frontier for functional ecology the incorporation of broad suites of traits within mechanistic and statistical predictive models across diverse forests.

Read the paper here.