Summary
This study examined the relationship between vegetation patterns at three spatial scales and five plant-growth attributes that were suspected to affect competitive ability in 10 species collected from two hayfield plots of different ages (cultivated in 1974 and 1984). Relative abundance of species at the plot scale (30 X 100 m) showed no relationship with any of the attributes. Species diversity within 1 x 1 m quadrats, however, had a significant negative relationship with mean seed weight in the 1984 plot and a significant negative relationship with mean plant height, mean lateral spread, mean growth rate, and "potential growth index" (combining all five attributes) in the 1974 plot. Multiple regression analysis indicated that the relationship between the five plant attributes and species diversity within the 1 x 1 m quadrats was stronger for the older (1974) (r2 = 0.67) than for the younger plot (r2 = 0.32). These data suggest that (i) the role of "competitive ability" attributes in affecting community structure becomes more evident in later stages of community development, (ii) different attributes may affect competitive ability at different stages of grassland community development, and (iii) the consequences of attribute differences among species may be revealed only in patterns at the local neighbourhood scale within vegetation.
Methodology
Study site
The study site was located in a hayfield at the Queen's University Biological Station in Frontenac County, Ont. In 1974, the hayfield was ploughed, tilled, and sown with a hay mixture of Phleutn pratense L., Trifolium pratense L., and Medicago sativa L. (10, 5, and 5 kglha, respectively) (G. Hughson, personal communication). Since 1974 the field had been cut for hay annually in July. In 1984, two rectangular plots, 30 x 100 m, were demarcated within the hayfield. The two plots were chosen such that both were relatively flat and similar in vegetation composition. In June 1984, one of these plots, chosen randomly, received a treatment similar to that experienced by the hayfield 10 years before. The plot was ploughed, tilled, and sown with the same species mixture. We refer to this plot as the 1984 plot. The second field plot was left undisturbed in its 10-year-old state. This is referred to as the 1974 plot. During the study, hay cutting in the two field plots ceased; the surrounding field, however, was cut annually. To eliminate edge effects in each plot, a border zone of 5 m along the sides and 10 m at each end was excluded from sampling.
Several growth attributes of adult plants were measured in an experimental field plot situated 500 m south of the above field plots in a similar hayfield. An area 20 x 30 m was ploughed, tilled, and raked level in the spring of 1985. The plot was fenced along its perimeter to exclude large herbivores.
Field surveys
In July 1985, the two field plots were surveyed to determine the association of species at two different scales. Within each plot, 50 randomly positioned 1 x 1 m quadrats were surveyed for percent cover of species using the point-intercept method. At 100 regularly spaced (10 cm apart) points within the quadrat, a drill rod, 2 mm in diameter, was dropped perpendicular to the soil surface. For each sample point, all plant species that contacted the drill rod were recorded. Percent cover for each species was then calculated as the proportion of points (out of 100) at which the species was recorded within each quadrat. Within each of the 1 x 1 m quadrats, two 5 X 5 cm quadrats were randomly placed. This smaller quadrat was the smallest square quadrat size that contained on average only two species based on a preliminary survey of different-sized quadrats. Percent cover of species within the small quadrats was estimated visually to the nearest 5%.