A direct comparison of ecological theories for predicting the relationship between plant traits and growth

Abstract

AbstractDespite long‐standing theory for classifying plant ecological strategies, limited data directly link organismal traits to whole‐plant growth rates (GRs). We compared trait‐growth relationships based on three prominent theories: growth analysis, Grime's competitive–stress tolerant–ruderal (CSR) triangle, and the leaf economics spectrum (LES). Under these schemes, growth is hypothesized to be predicted by traits related to relative biomass investment, leaf structure, or gas exchange, respectively. We also considered traits not included in these theories but that might provide potential alternative best predictors of growth. In phylogenetic analyses of 30 diverse milkweeds (Asclepias spp.) and 21 morphological and physiological traits, GR (total biomass produced per day) varied 50‐fold and was best predicted by biomass allocation to leaves (as predicted by growth analysis) and the CSR traits of leaf size and leaf dry matter content. Total leaf area (LA) and plant height were also excellent predictors of whole‐plant GRs. Despite two LES traits correlating with growth (mass‐based leaf nitrogen and area‐based leaf phosphorus contents), these were in the opposite direction of that predicted by LES, such that higher N and P contents corresponded to slower growth. The remaining LES traits (e.g., leaf gas exchange) were not predictive of plant GRs. Overall, differences in GR were driven more by whole‐plant characteristics such as biomass fractions and total LA than individual leaf‐level traits such as photosynthetic rate or specific leaf area. Our results are most consistent with classical growth analysis—combining leaf traits with whole‐plant allocation to best predict growth. However, given that destructive biomass measures are often not feasible, applying easy‐to‐measure leaf traits associated with the CSR classification appear more predictive of whole‐plant growth than LES traits. Testing the generality of this result across additional taxa would further improve our ability to predict whole‐plant growth from functional traits across scales.