• Community aggregated traits;
  • Community weighted traits;
  • Deciduous forests;
  • Plant functional traits;
  • Variance components;
  • Variance decomposition;
  • Within-species variability;
  • β-diversity



When can we assume that inter-specific trait variation is higher than intra-specific trait variation in plant community ecology?


Old-growth deciduous forest in the Gault Nature Reserve, Mount St. Hilaire (Quebec, Canada,45′32.957″ N, 73′08.884″ W), with a shorter environmental gradient than normally exists in community studies.


We measured 15 functional traits on all tree saplings occurring in 39 (5–7-m radius) sample plots. Using variance decomposition from mixed models and from sums of squares of community-weighted traits, we determined the relative importance of temporal, spatial, inter-specific, intra-specific and intra-individual variation in each trait. In total, we collected trait information on 3317 leaves from 786 twigs sampled on 422 saplings.


For 12 of 15 traits, over 50% of the total variance existed between species and inter-specific variation was always the most important source of variation. However for 14 traits, intra-specific and environmental variation represented up to 28% of the total variation. Variation in community-weighted trait means was mostly generated by changes in species composition, but intra-specific trait variation was the dominate cause for leaf nitrogen, specific leaf area and tree branching. The intra-individual and temporal sources of variation were not important.


Trait variation should be dominated by inter-specific differences in most studies since they involve systems with more pronounced environmental gradients, higher species richness and more β-diversity than used here. However, in studies with short gradients, or when using more plastic traits, it will be necessary to measure trait values for each site. It is important to quantify the β-diversity of the environmental gradients in question in order to compare results across studies.