The statistical analysis of barley quality traits – Tiers for Beers
by
Brian Cullis and Patrick Lim
NSW Agriculture, Wagga Wagga, and BiometricsSA, The University of Adelaide
Coauthors: Alison Smith (Wagga Wagga Agricultural Institute, Pine Gully Road, WAGGA WAGGA NSW 2650), Joe Panozzo (Agriculture Victoria, Horsham, PMB 260, HORSHAM VIC 3401)

Although there have been many advances in the design and analysis of yield data in plant improvement programs in Australia there has been little progress made in routine analysis of quality traits. The measurement of most quality traits involves a complex multi-phase experiment in which there is a field phase and several laboratory phases. Quality trait data is rarely subjected to a proper statistical analysis. Routine selection in most plant improvement programs is based on tabulation of raw means across one or more environments of composite field samples. Generally there is no replication at the field phase and most laboratories only process laboratory checks at a very low frequency. The efficiency of this approach is unknown as there is little knowledge of the relative magnitude of the sources of variation in the field and laboratory. Furthermore, since most quality traits are linked to grain protein which is linked to yield it is quite likely that spatial heterogeneity may be present in the field phase. The selection for quality is a key component of the selection process and it is therefore crucial that resources be allocated to maximise genetic gain.

In this talk we consider the analysis of barley malting trait data taken from a subset of the National Barley Molecular Marker program. Since virtually all field plots were sampled and partially duplicated in the laboratory the opportunity existed to examine the sources of variation for these types of data. The talk is in two distinct parts. In the first part we present a thorough description of the malting process and the experiment design. In the second part we develop the linear mixed model for a randomisation based analysis using the concept of tiers (Brien, 1983) and then extend this model to accommodate important sources of heterogeneity. For example, phase one consists of a series of trials, also known as a multi-environment trial (MET). Smith et al (2001) present an approach for modeling genotype by environment interaction in the analysis of grain yield data from METs. We show how their approach can be integrated with the randomisation based analysis for quality trait data to account for spatial dependence in both phases of the experiment.

References: Brien, C. J. (1983). Analysis of variance tables based on experiment structure. Biometrics, 39: 53-59.

Smith, A.B., Cullis, B.R. and Thompson, R. (2001). Analysing variety by environment data using multiplicative mixed models. Biometrics, 57: 1138-1147.

Smith, A.B., Cullis, B.R., Appels, R., Campbell, A.W., Cornish, G.B.,Martin, D. and Allen, H.M. (2001) The statistical analysis of quality traits in plant improvement programs with applications to the mapping of milling yield in wheat. Australian Journal of Agricultural Research 52, 1207-1219.

Date received: October 13, 2002

Copyright © 2002 by the author(s). The author(s) of this document and the organizers of the conference have granted their consent to include this abstract in Atlas Conferences Inc. Document # cajn-53.