Summary
Large plant species self-thin to disproportionately lower densities than smaller plant species, and therefore may leave more patches of unused space suitable for invasion. Using experimental monocultures of 11 old-field perennial plant species differing in maximum size, as well as mixtures composed of all monoculture species, we tested our primary hypothesis that monocultures of larger species will be more susceptible to natural invasion. After 3 years, monocultures of larger species were invaded by a significantly greater number of species, and more ramets, from the surrounding vegetation. Invading plant species were significantly smaller than the monoculture species being invaded, suggesting that smaller plant species may be better invaders. Thus, we quantified a trade-off between species size, which is frequently associated with increased competitive ability for light, and invasibility, suggesting one reason why large and small species coexist in virtually all plant communities. Although we expected that invasion would enhance biomass production by more fully capturing available resources, we found that the most highly invaded plots of each species produced significantly less biomass. This suggests that increased diversity resulting from invasion did not result in complementary resource use. Mixture plots containing all experimental species did not admit a significantly different number of invading ramets or species than most monocultures, indicating no obvious role for diversity in resistance to invasion, or complementary resource use. Our results suggest that relatively large species may be limited in their capacity to competitively exclude other, smaller species from communities because pure stands of the former are more susceptible to invasion by the latter.
Methodology
The study was conducted at the Queen’s University Biology Station, Canada (44°34′N, 76°20′W), in a 30 × 30-m section of an old-field. In April 2004, prior to the start of the experiment, the field was tilled to 30 cm, and again several times to 10 cm to disrupt the growth of rooted plants and to homogenize the soil.
A total of 226 treatment plots (40 × 40 cm) were set up in a grid (15 rows × 16 columns of plots) with 1-m separation between plots. Edge plots were 2 m from the surrounding fence. Each plot was delineated by a wooden frame 10 cm high and buried to a depth of 5 cm.
Plots were seeded as either monocultures or mixtures. Monocultures of 17 species of herbaceous perennial plant species were seeded into 166 plots, with ten replicates for all but one species, which had only six replicates (this species was later eliminated from analysis—see below). Target species were chosen to span the existing species size distribution in a nearby old-field plant community (Fig. S1a), although this correspondence was slightly reduced with the loss of some monoculture species (Fig. S1b; see below for explanation). Treatments were assigned to plots according to a stratified random design, so that species were evenly represented among rows and randomly placed within columns. The number of invading ramets and species, and plot biomass did not vary significantly by row; although there was a column effect on invasion (see Results). The remaining 60 plots were seeded as mixtures of the 17 species such that total plot density was expected to equal that of monocultures, with equal numbers of each species (i.e., each species making up 1/17th of the community). Grazing by deer killed most individuals in monocultures of six species at the end of the first growing season. These plots were excluded from the analysis; the remaining 11 monoculture species still spanned a wide range of maximum aboveground biomass (Table 1).