(1) The sensitivity of replacement-series experiments to variations in density was investigated in Agropyron repens, Phleum pratense and Poa pratensis collected from a hayfield in south-eastern Ontario, Canada. All pairwise combinations of species were grown in a replacement-series design at three different densities, in four pot sizes, for three experimental durations.
(2) Analysis of variance indicated that monoculture pot yields, yield suppression ratio (YSR) and relative yield total (RYT) were sensitive to density, experimental duration and pot size. The density dependence of YSR was, however, restricted to experimental conditions where component yield was below 'constant final yield'.
(3) The results suggested that interpretations from replacement-series experiments are not sensitive to variations in density per se, but to the extent to which the supply of resources exceeds the demands made on them. This may be controlled for by choosing a mixture density (N) such that each component yield (at density N/2) would be at constant final yield if in a monoculture at that density.
Seeds of Poa pratensis L., Phleum pratense L. and Agropyron repens L. Beauv. (nomenclature follows Gleason & Cronquist 1963) were collected in 1986 from a hayfield at the Queen's University Biological Station (44°34'02"N, 76°21'52"W). These species were the three most abundant species in the field, with 32.8%, 27.2% and 12.2% cover, respectively.
Seeds were germinated on flat trays moistened with a 0.1% giberellic acid solution in growth chambers with temperatures fluctuating diurnally between 10 °C and 25 °C. P. pratense germinated most quickly and uniformly and P. pratensis took longest to germinate. To facilitate the simultaneous germination of all three species, therefore, the onset of exposure of the seeds to the giberellic acid solution was staggered. P. pratensis and A. repens were sown in the growth chamber ten days before planting, while P. pratense was sown only four days before planting.
Ten days after the first species was sown, seedlings were planted in pots filled with a standard potting mixture of peat and perlite in a temperature-controlled glasshouse. Each of the three species was planted in monoculture at five densities (1, 2, 5, 10 and 20 plants pot-1). Each pairwise combination of species was planted in an equiproportional mixture at three densities (2, 10 and 20 plants pot-1). Mixtures and monocultures at each density were grown in pots of four different sizes (5.5, 7.3, 10.2 and 15.0 cm top-diameter standard plastic pots) and for three different durations (16, 20 and 24 weeks). There were three replicates of each monoculture and five replicates of each mixture.
All pots were saturated with water every two days. Two weeks after planting, dead plants were replaced with surplus seedlings of the same species to obtain the initial densities described above. All pots were repositioned randomly every two weeks. After eight weeks, plants exhibited signs of nutrient deprivation. Subsequently, all pots were fertilized at four-week intervals with 0.10 pot-1 of Plant Prod 20-20-20 fertilizer (Plant Products Ltd, Brampton, Ontario). At each harvest, the total above-ground dry weight was recorded for each species in each pot.