• Schamp, Brandon S.
  • Aarssen, Lonnie W.


An ongoing debate in ecology concerns the relative importance of competition in driving community patterns, especially along gradients of disturbance and productivity. We used a null model to address this question by testing for non‐random organization of forest species according to maximum height, a trait associated with competitive ability for light. Specifically, we compared the species present in 409 different temperate forest plots with the pool of potentially present species (n=639 species), spanning a 50 000 km2 area of southern Ontario, Canada. In contrast to current theory, coexisting forest species were neither more similar (i.e. convergent), nor more different (i.e. limiting similarity) in maximum height than expected by chance. However, coexisting forest species had larger maximum heights on average, and their maximum heights were more symmetrically distributed than expected by chance, suggesting that competition has reduced the representation of smaller plant species within plots (i.e. higher turn‐over of species with smaller maximum heights among forest plots). We explored the possibility that our findings resulted from smaller species having relatively narrower niches; however, a conclusive test of this explanation will require knowledge of fundamental, rather than realized niche breadth. We also tested the prediction that the influence of competition changes along gradients of productivity and disturbance by examining how the effect size of our null model tests changed along these gradients. We observed that species with smaller maximum heights were increasingly under‐represented in more productive forest communities, suggesting an increased role for competition in determining species membership in more productive communities. In contrast to theory, however, the effect size of our tests did not significantly change along a gradient of forest disturbance. In summary, we found evidence that maximum species height plays a significant role in driving the non‐random organization of plant species among hundreds of mature forest plots, and that this role is more pronounced in more productive forest plots.


Data were collected as part of the Ecological Land Classification (ELC) project for southern Ontario forests undertaken by the Ontario Ministry of Natural Resources (Lee et al. 1998, Schamp et al. 2003). Data were collected from mature natural forests (>50 years old) spanning the range of landforms, soils and topographies in the region and therefore capture the full extent of gradients of both productivity and disturbance (n = 409). The study spans a 50 000 km2 region of deciduous and mixed coniferous deciduous forests in southern Ontario. The most common tree species are Acer saccharum, Fraxinus americana, Acer rubrum, Prunus serotina, Ostrya virginiana, Tilia americana and Quercus rubra. The most common herbaceous species include Maianthemum canadense, Epipactis helleborine, Arisaema triphyllum, Polygonum pubescens and Maianthemum racemosum.

In each forest plot, all vascular plant species within a 1010 m square were identified; all forests were sampled between 1996 and 2003. This plot size was chosen to be large enough to represent the forest community, but small enough to reduce the influence of habitat heterogeneity (Lee et al. 1998). To remove the potential for bias that results from taller species occurring in a plot but only as small plants, we only listed tree species as present in a plot if they had reached the sub-canopy or canopy layer. In some cases, the ELC dataset consisted of multiple plots in the same forest; only one plot per forest was included in the analysis to set a baseline for statistical independence. Also, because forests varied greatly in size and connectivity, the criteria of plots being in separate forests was sometimes unclear.

Therefore, plots used in analyses were limited to those that were greater than one kilometer apart. To obtain species-specific trait data, we collected data on maximum species height for each of the 639 species observed in our forests data from published floras (Gleason and Cronquist 1991). For tree species, we used maximum height data collected from our study region as part of the ELC project. The maximum potential height for tree species was the maximum height per species observed from 5900 field measurements of 67 tree species.