Ecological Archives E096-255-A5
Geneviève Lajoie and Mark Vellend. 2015. Understanding context dependence in the contribution of intraspecific variation to community trait–environment matching. Ecology 96:2912–2922. http://dx.doi.org/10.1890/15-0156.1
Appendix E. Statistical details on testing the impact of trait-abundance relationships within species (Question 3C) and variation in intraspecific trait-environment slopes (Question 3D) on the contribution of ITV to trait-environment matching.
Statistical details on testing the impact of trait-abundance relationships within species (Question 3C) and variation in intraspecific trait-environment slopes (Question 3D) on the contribution of ITV to trait-environment matching
1. Testing the impact of trait-abundance relationships within species
To test for the impact of trait-abundance relationships within species on the contribution of ITV to trait-environment matching along the elevation gradient (Question 3C, see Fig. 1 in the main text), we used random permutations of species raw abundances. In each permutation, we shuffled raw abundances of each species among the sites where it was already present. The trait values were kept intact, so as to break intraspecific relationships between trait and abundance distributions. We then calculated the relative contribution of ITV (%ITV) using this new abundance matrix, performing the same calculations as before. We calculated the proportion of simulated values that were lower than the observed (Psim<obs) using 9999 permutations, so as to assess whether trait-abundance relationships contributed significantly to determining %ITV (see also Jung et al. 2014).
2. Testing the impact of species slope distribution
To assess the influence of variation in the direction of intraspecific trait-environment slopes on the detected contribution of ITV to community trait-environment matching (Question 3D, see Fig. 1 in the main text), we calculated theoretical maximum %ITV values for SLA and height. In the case of a trait whose community level average varies monotonically across the gradient, the expectation is straightforward: ITV contributes most when intraspecific slopes are consistent across species, aligning with the community-level pattern. For quadratic community-level trait-environment relationships, the expectation is less straightforward: ITV contributes most when intraspecific slopes align with the community-level pattern, but only if SPT and ITV covary in all parts of the gradient. Because this assumption did not hold in one trait (height), we addressed Question 3C conservatively by altering intraspecific slopes such that they would all be of the same direction as the most commonly observed response to the gradient (as described below) and then recalculating %ITV as the theoretical maximum. With essentially all intraspecific flowering time-elevation slopes being positive, this analysis was therefore not conducted for this trait.
Prior to altering slopes, intraspecific trait-environment relationships were modelled with both linear and quadratic terms for each species recorded in at least 4 plots along the gradient. The sign of the linear component of the best fit model (identified using F test) was extracted to determine the direction of species slope. Slope reversal was performed by reversing the order of trait values along the elevational gradient. For example, for a species having the following trait distribution in four sequential plots along the mountain slope (2,5,4,9: positive slope with elevation), the trait values would now be reversed to (9,4,5,2: negative slope). We thereby reverse the slope sign while maintaining the distribution of trait values. Once the new trait matrix was generated, we calculated %ITV using the same procedure described in the main text.
Literature cited
Jung, V., C. H. Albert, C. Violle, G. Kunstler, G. Loucougaray, and T. Spiegelberger. 2014. Intraspecific trait variability mediates the response of subalpine grassland communities to extreme drought events. Journal of Ecology 102:45–53.