Assessment of five cultivars of potato (Solatium tuberosum L.) in a competition diallel

In 1992 and 1993, potato cultivars Cara, Estima, Pentland Dell, Pentland Javelin and Torridon were assessed in all possible pairings in three consecutive drills, such that one member of the pair (the cultivar) was the centre-drill, and the other member (the neighbour) was in the two adjacent drills. Each year the trial was a randomised complete block design with four replicates. Inter-cultivar competition was present for height, total tuber yield, ware yield and dry matter content, but not for emergence and maturity. Cultivar heights and dry matter contents under inter-cultivar competition and in pure-stands were highly correlated, whereas for total and ware yields, the rankings of cultivars in pure-stands were different from those under competition. Pure-stand yields were predicted for the two years separately because of year × cultivar and year × neighbour interactions. When the average total yield of the outer-drills was used as a covariate to adjust centre-drill yields to pure-stands, predicted yields were closer to those observed than were yields under competition, but there was little improvement in the correlations. In contrast, not only were the predicted values even closer to the observed ones when based on individual neighbour effects, the correlations were much higher (r = 0.76–0.94). Unfortunately, in plant breeding trials with large numbers of entries, precise estimation of individual neighbour effects would require an unrealistically large number of replicates. Hence, for the final stages of yield assessment of new cultivars, it is still desirable to use multiple-drill plots and to discard the outside drills where inter-cultivar competition can occur.