Theory has produced contrasting predictions related to flowering time overlap among coexisting plant species largely because of the diversity of potential influences on flowering time. In this study, we use a trait‐based null modelling approach to test for evidence of deterministic assembly of species according to flowering time in an old‐field plant community.
Plant species coexisting in one‐metre‐square plots overlapped in flowering time significantly more than expected. This flowering synchrony was more pronounced when analyses focused on bee‐pollinated species. Flowering synchrony was also observed for wind‐pollinated species, although for only one of our two null model tests, highlighting the sensitivity of some results to different randomization methods. In general, these patterns suggest that relationships between pollinators and plants can influence community assembly processes.
Because our study community is composed of approximately 43% native plant species and 57% exotic species, and because the arrival of new species may complicate plant–pollinator interactions, we tested whether flowering time overlap was altered by introduced species. Flowering synchrony was greater in plots with a higher proportion of introduced species. This pattern held for both null model tests, but was slightly stronger when analyses focused on bee‐pollinated species. These results indicate that introduced species alter community flowering distributions and in so doing will inevitably affect pollinator–plant interactions.
Finally, we tested whether our results were influenced by variation among study plots in above‐ground biomass production, which some theory predicts will be related to the importance of competition. Our results were not influenced by this variation, suggesting that resource variation among our plots did not contribute to observed patterns.
Synthesis: Our results provide support for predictions that coexisting species should display flowering synchrony, and provide no support for species coexistence via temporal niche partitioning at this scale in this study community. Our results also indicate that introduced species significantly alter the community assembly process such that flowering synchrony is more pronounced in plots with a greater proportion of introduced plant species.
This research was conducted in an old‐field plant community from May to October 2009 at the Queen's University Biological Station, Chaffey's Locks, Ontario, Canada (44°34′N, 76°20′W). The field is relatively homogeneous topographically and hosts 37 plant species known to differ considerably with respect to multiple functional traits (Schamp, Chau & Aarssen 2008) including flowering time (Fig. 1). The study field has not been ploughed or tilled for more than 40 years, but in most years since then it has been mown for hay once per year. Vegetation in this community is dominated by perennial plant species, the most common of which include Poa pratensis, Phleum pratense, Cerastium arvense, Potentilla recta, Vicia cracca and Rumex acetosella.
A 50 × 50 m study area near the centre of the field was chosen to reduce any edge effects from the surrounding forest, and 50 1 × 1 m plots were randomly located within a 20 × 20 grid of possible plots. This plot size was chosen to reflect a scale at which plant species interactions, particularly negative interactions between neighbouring plants, are likely to be relevant. All 50 plots were surveyed early in the season (15 May 2009) to determine species composition and surveyed three additional times throughout the growing season (29 May, 17 July, and 9 October) to ensure that plot censuses were complete and accurate. Above‐ground biomass from each plot was harvested in October 2009, dried at 60°C for 3 days and then weighed. Also, 25 individual plants (i.e. ramets) of each species were located in the field, and marked at the base using a labelled aluminium tag (925 tagged plants). All tagged plants were located outside of the 50 study plots, but within the study field. Tagged individuals of each species were monitored once weekly from the start of May to the end of October in 2009 (24 weeks) to produce species‐specific flowering time distributions; not all individuals of a particular species were flowering in each week. Flowering was considered started when the first flower was observed to be open on an individual plant (Pleasants 1980), and flowering was considered terminated when the plant no longer possessed any flowers with anthers. All study species flowered within the 24 weeks of monitoring (Fig. S1, Supporting information). At the time monitoring began, some individuals of six species had already begun flowering. We ran sensitivity analyses, extending the flowering distribution start times for these species backwards up to 4 weeks based on information from published floras (Gleason & Cronquist 1991; Fitter & Peat 1994), to test whether missing the start of the flowering period for these species influenced the results.