Using mixing ability analysis from two-way cultivar mixtures to predict the performance of cultivars in complex mixtures

Cultivar mixtures are an alternative to monoculture for crop production. Me thods for predicting the performance of cultivars in mixtures would facilit ate the identification of the best cultivars for mixture formation. A predi ction method [Theor. Appl. Genet. 81 (1991) 265] based on the combining abi lity analysis proposed by Gardner and Eberhart [Biometrics 22 (1966) 439] m odel II was evaluated for mixtures. Mean values for percent diseased leaf a rea (DLA) and yield under disease (YUD) of five club wheat cultivars and al l possible two-way mixtures (across three Oregon locations) were used to es timate the relative contribution of each cultivar to the mixture mean (I-i) and the predicted mean of all possible mixtures (Y-m) according to the MFC method. Estimated I-i and Y-m were compared to actual mixture means, I-i a llowed identification of the best cultivars for mixture formation for both DLA and YUD. Actual and predicted rank correlation coefficients for DLA and YUD of complex mixtures (more than two components) were highly significant (P < 0.01) for the mean of the three environments (0.87 and 0.78, respecti vely) and for the mean of the two most relevant environments (0.83 and 0.93 , respectively). Similar results were obtained when I-i and Y-m were estima ted from the means of the cultivars in pure stand (instead of being estimat ed from the mixing ability analysis). This was due to the relatively small competitive effects (h(i)) of the cultivars compared to their additive effe cts (v(i)). We extended our analysis to data from a yield study of two-way mixtures of eight soybean cultivars [Crop Sci. 29 (1989) 885; Agron. J. 81 (1989) 559], where additive and competitive effects had similar magnitude. Actual and predicted mixture means could not be compared because only two-w ay mixtures were included in the study. In this case, the mean yield of the cultivars in pure stand was not a good predictor of the performance of cul tivars in mixtures. Ra 604, the third highest yielding cultivar, had a posi tive additive effect (v(i) = 53.8 kg/ha), but a highly negative competitive effect (h(i) = -49.7 kg/ha) that resulted in a negative contribution (I-i) to yield when mixed with the other soybean cultivars. Additive (v(i)) and competitive effects (h(i)) must be considered to obtain superior mixtures, and the advantage of I-i is that it takes into account both effects (I-i = 1/2v(i) + (k - 1/k)h(i), k is the number of cultivars). The MFC method may be a useful tool to select desirable cultivars to obtain complex mixtures. (C) 2000 Elsevier Science B.V. All rights reserved.