Published online by Cambridge University Press: 27 March 2009
Four trials were carried out to investigate the effect of inter-plot competition on seed yield and plant height of field beans using experimental designs balanced for neighbours. One trial tested four varieties ranging in final height from 25 to 100 cm and sown in four-row plots with 30 cm spacing between rows and one blank row between plots. When grown between plots of a dwarf variety, the plot yield of the tallest variety was increased by 20% compared with its pure stand yield: a complementary reduction in yield was shown by the dwarf variety when grown between plots of the tall. The effect on the yield of the inner rows of the plot was somewhat smaller than that of the outer rows but still significant. Another trial with six commercial varieties differing in final height by less than 15 cm showed no differential effects of inter-varietal competition on yield. In two trials with six varieties of widely differing heights planted with 50 cm spacing between rows, but no gap between plots, plot yield was strongly affected by the neighbouring variety in the trial with single-row plots, while yields from the four-row trial showed a large interaction between variety and inner and outer row position, again indicative of inter-variety competition. The yield response to competition could be described by a common linear regression on the excess height of plot neighbours: there was little evidence of varietal differences in sensitivity or aggressiveness.
Plant height showed a positive response to height of plot neighbours in all four trials, whilst in the single-row trial, the ratio of grain yield to total dry-matter production (harvest index) was negatively related to excess height of neighbours.
The results suggest that, when testing varieties differing in height by more than 20 cm, at least two buffer rows, e.g. six-row plots with the centre two rows harvested, are required to obtain a true comparison of pure-stand yield. Alternatively, when this procedure would result in an unacceptable use of resources, one of the statistical models adapted in this paper may be used to adjust variety yields for pure stands.