A key process determining community assembly is habitat filtering, which reduces the range of trait values and thus leads to trait convergence. Habitat filtering is often assumed to involve a reduction in species richness, but such trait convergence could also result from phenotypic plasticity. By allowing more species in the community to show the “right” phenotype and pass the filter, phenotypic plasticity could maintain species richness despite trait convergence. If phenotypic plasticity plays this role, environmental changes should modify trait distribution but not species composition or richness. We tested this hypothesis in a forest edge community recently invaded by the alien shrub Rubus praecox. We measured functional traits beneath Rubus canopy (Rubus-invaded community) and in adjacent microsites naturally free from Rubus (original community), comparing trait dispersion and community structure parameters. We also characterized the microenvironments to seek functional explanations for the observed phenotypic changes. The functional dispersion parameter FDis showed habitat filtering in the Rubus-invaded community (trait convergence), while in the original community it did not differ from a null model. Species richness and composition did not differ between the two communities. Soil characteristics were similar in both communities but light availability was greatly reduced in the Rubus-invaded community. A Principal Component Analysis of community-weighted means of functional traits showed that phenotypic changes between communities were mostly consistent with functional responses to shading, including an increase in specific leaf area and a reduction in flavonoid concentration in the Rubus-invaded community. We verified the conditions necessary to infer that phenotypic plasticity mediates the outcome of habitat filtering driven by Rubus in the study system, i.e., plasticity would explain species persistence despite trait convergence. Phenotypic plasticity, often subsumed within intraspecific variation in community assembly studies, should be considered individually to improve our understanding of the causes and consequences of habitat filtering.

Trait Data – Functional traits measured in each block, per species, in both original and adjacent Rubus-invaded communities. Trait abbreviations as indicated in the text and in the legend of Figure 4. n/a: trait not measured.

Community Data – Species presence and abundance in each block, in both original and adjacent Rubus-invaded communities.

Soil Data – Soil parameters measured in each block, per species, in both original and adjacent Rubus-invaded communities. Trait abbreviations as indicated in the text and in the legend of Figure S1.

Light Data – Photosynthetically active radiation (µmol m^-2 s^-1) 5 cm above ground level in each block, in both original and adjacent Rubus-invaded communities.