Multiplicative main cultivar effects in Czech official winter wheat trials 1976-2005

Abstract

Abstract:98 winter wheat cultivars, tested by the Czech Plant Variety Office in 554 trials between 1976 and 2005, were analysed for breeding progress, yielding capacity and behaviour in particular environments. The trials were grouped by yield level, years, altitude, precipitation, temperature and ecological regions. Within each group, least squares (LS) adjusted cultivar means and variance components were estimated. The cultivar means within groups were used as primary data for the subsequent analysis across groups. Genetic variance (Vg) increased with increasing yield level while residual variance (Vr) slightly decreased. Although Vg within groups was very different, phenotypic correlations between the cultivar means of the groups were very high. The cultivar yields, plotted against environment yield, followed a funnel-like distribution, expanding from the origin. The data fitted better to the assumption of multiplicative main cultivar effects than to the assumption of additive main cultivar effects, implied by standard models. Standardised relative yields (SRY), assuming multiplicative main cultivar effects on the original scale, were therefore used to analyse the data. The SRY were obtained by standardisation of cultivar means within trial groups to the variance expected from the cultivar means across the trial groups, subtraction of additive environment effects and division by the general mean. The SRY had a smaller relative error, as measured by the Vg/Vr relation, than cultivar means calculated under the assumption of additive main cultivar effects. The SRY were surprisingly consistent over environments. G × E interactions, usually found under the assumption of additive main cultivar effects, may partly be artefacts of the assumption.

Yield stability and resistance to leaf spot diseases and rosette in groundnut

“…It is shown as the last column in Table 4. We are aware that stability parameters, calculated from a limited number of environments, are subjected to enormous errors (Piepho 1998), are difficult to reproduce (Eagles & Frey 1977) and may be partly artefacts of inappropriate models (Schwarzbach et al 2007). Therefore we tried also the multiplicative model, which fits the natural behaviour of some crops better than the additive model.…”

“…A two-way analysis of variance (ANOVA) was performed and from the expectations of mean squares estimates of variance components were calculated. Two statistical models were used: The a i estimates in (b) were calculated as described by Schwarzbach et al (2007), by standardization of the data through division by the local standard deviation, calculation of location effects and subtraction of these from the standardized data. The matrix of the a i estimates has then zero location effects and their variances within environments are homogeneous.…”