Methods of analysis of a competition diallel in Italian ryegrass with varying mixture proportions

Abstract

SUMMARYTwo methods of analysis of variance of experiments involving diallel arrangements of binary mixtures and monocultures where the former are each represented by a linear replacement series (de Wit, 1960) to detect mean quadratic and quartic patterns among the monocultures and different mixture types, and array differences with respect to linear, quadratic, cubic and quartic effects are given. These are applied to data from an experiment on Italian ryegrass where the four varieties 1145 and S. 22 (diploid), IG2 and Tetila (tetraploid) are mixed in the proportions 75:25, 50:50 and 25:75.Significant mean quadratic effects for D.M. yield at certain harvests are due to higher yields from monocultures than mixtures, and a mean quartic effect at one harvest reflects a lower yield from 75:25 mixtures than 50:50 mixtures or monocultures. Array differences for linear and quadratic effects are largely a property of the ploidy contrast. Specific mixture effects were recognized for two varietal combinations, mixtures between the two tetraploid varieties showing some yield depression, and those between the two commercial varieties (RvP and Tetila) a considerable increase. No meaningful significant effects were detected for in vitro digestibility.Further analysis of total D.M. yield by means of the regression of array covariance (Wr) on to array variance (Vr) show, as expected, a closer relationship of Wr and Vr with the non-recurrent component at a frequency of 75% than at 50% or 25%, but the significant departure of the slope from unity is interpreted as evidence that no reliable information as to the relative competitive abilities of the varieties can be obtained from these statistics. Examination of the de Wit replacement graphs strongly suggests a ranking of competitive ability which accords with the known morphology of the varieties.