Isoyield Analysis of Barley Cultivar Trials in the Canadian Prairies

Abstract

Classification of test sites used for cultivar trials into groups with similar within-group site performance and response (isoyield groups) is an important step towards identification of appropriate cultivars that are best suited for different productivity levels in farm fields. The objective of this study was to determine isoyield environments in the Canadian prairies based on the analysis of cultivar trials consolidated from individual provinces for barley (Hordeum vulgare L.). Yield data for the analysis were taken from 324 replicated trials at 84 sites across the prairies during 1995-2003. The combined use of regression and cluster analyses of the data normalized for averaging the multi-year unbalanced data led to a stratification of the 84 sites into 13 isoyield groups. A comparison was made of the distributions of the variability among and within groups according to three modes of grouping: isoyield groups, soil zones and agroecoregions. There was more variability among isoyield groups and correspondingly less within the groups than that among and within soil zones or agroecoreions. Similar contrasting pattern existed for the variance components involving genotype-environment interaction (GEI), although the GEI variability was generally small under all three modes of grouping. Relationships of site sensitivity (regression coefficient) and stability (coefficient of determination) with site productivity were shown to be a useful aid for selecting a subset of test sites in an effort to improve efficiency and quality of future cultivar testing. Thus, isoyield analysis should be a valuable tool for subsetting heterogeneous environments and for reducing GEI impact in cultivar testing and recommendation.

Wheat yield in Spain and associated solar radiation patterns

“…Our classification considered only wheat variability, which is different from the method used by Iglesias et al (), who applied cluster analysis to group Spain into agro‐climatic regions based on temperature and precipitation data. Other agro‐climate classifications have considered yield in combination with topographical or soil characteristics (Yang et al , ).…”

“…Our classification considered only wheat variability, which is different from the method used by Iglesias et al (2000), who applied cluster analysis to group Spain into agro-climatic regions based on temperature and precipitation data. Other agro-climate classifications have considered yield in combination with topographical or soil characteristics (Yang et al, 2006). K-means clustering aims to partition a given dataset into a number of clusters in which each cluster is determined by the degree of similarity and difference between observations.…”

Biplot Analysis of Genotype × Environment Interaction: Proceed with Caution

“…Since unpredictable year-to-year weather fl uctuation typical in the Canadian Prairies contributes signifi cantly to yield variation and site instability, Yang et al (2005Yang et al ( , 2006 suggested a normalized procedure to fi lter out much of the random variation among years so that clustering locations would be based on more accurate estimates of location averages. However, it remains largely unknown as to the eff ectiveness of such a normalization as a general procedure because it depends on the level of imbalance in year × location × genotype combinations and on yearly diff erences in location × genotype cell means.…”

Mixed-Model Analysis of Crossover Genotype-Environment Interactions

“…The fi rst data set used for the mixed-eff ect model, involving fi xed genotypic eff ects and random environmental eff ects, was taken from the Canadian Prairies Barley Trials as described in Yang et al (2006). The fi rst data set used for the mixed-eff ect model, involving fi xed genotypic eff ects and random environmental eff ects, was taken from the Canadian Prairies Barley Trials as described in Yang et al (2006).…”

“…Two data sets were used for the COI assessment under the mixed-eff ect model. The fi rst data set used for the mixed-eff ect model, involving fi xed genotypic eff ects and random environmental eff ects, was taken from the Canadian Prairies Barley Trials as described in Yang et al (2006). While these trials included a large number of cultivars and advanced breeding lines, a focus has been on the following six two-row barley cultivars (Helm et al, 2004): three eligible for feed grades-'CDC Dolly', 'Seebe', and 'Xena'-and three others eligible for malting grades-'AC Metcalfe', 'Harrington', and 'Merit'.…”