Variety Results Information

Field Methods

Randomized complete block designs with four replications were used at all trials locations, except the Intermountain regions, for which three replications were used prior to 2016. Tests were sown at seeding rates of approximately to 1.2 million seeds per acre for all tests (equivalent to 61 to 107 lbs/acre for common wheat, 78 to 99 lbs/acre for triticale, 75 to 140 lbs/acre for durum wheat and 77 to 113 lbs/acre for barley, depending on the variety). Each plot was six or nine drill rows wide (5 to 8-inch row spacing) and 15 to 20 feet long, target planted area being 100 ft2. Grain was harvested with a Wintersteiger Seedmaster Universal 150 plot combine.

Analytical Methods

Data Groupings

Genotype × environment effects the trial data were analyzed and interpreted using the gge package in R. The analyses of genotype × environment patterns across multiple seasons suggest that, for all the species tested, California can be subdivided into five sub-regions that may display different rankings of small grains varieties, particularly in the case of common wheat yield. These regions are the Inter-Mountain, Sacramento Valley, Northern San Joaquin Valley, Southern San Joaquin Valley, and the Imperial Valley. These sub-regions may change given new data or the introduction of new varieties. For an overview of how biplot analyses are applied to multi-environment crop trial data see Yan [1, 2].

Means

1) For site titled Choose a small grain variety based on UC trial data:

Within the sub-regions described above, the variety performance, in terms of yield and protein content, is summarized using a linear mixed model approach via the lme package in R, with yield and protein expressed as lsmeans across locations for the most recent 3 seasons in a given grouping. Linear mixed models have advantages for analyzing data from multi-environment trials as compared to ordinary linear models such as analysis of variance (ANOVA). These include the ease with which incomplete data (i.e. not all varieties in all years and locations) can be handled, the ability to use more realistic within-trial models for error variation (e.g. spatial correlation models), improved means estimates by better ascribing variance to different experimental factors, and the ability to assume some sets of effects (e.g. trial location) to be random rather than fixed. Multi-year summaries are more appropriate for assessing variety performance than single year summaries.

2) For the site titled Explore simple summaries of multi-year small grain variety trials in California: 

Means, standard deviations and an average coefficient of variance with respect to yields (i.e. trial mean of the CV for each variety) were calculated. 

Disease and Agronomic Traits

Foliar diseases were assessed at the soft-to-medium dough stage of growth by estimating the percentages of areas of penultimate leaves (flag leaf) affected. Barley Yellow Dwarf Virus (BYDV) assessments, however, were based on the percentage of plants showing symptoms. Lodging and shatter observations were rated by the percentage of plants affected in the plot.

For all available trial data for a given variety, disease, lodging and shatter observations scores were sorted in ascending order, and the interpolated 90^th percentile value was calculated via the quantile() function in R. Using this approach, the disease and agronomic trait summaries highlight reactions for conditions under which disease, lodging and shatter were more likely present. However, varieties that have been tested for longer periods of time are more likely to demonstrate susceptibility.

For more information, please contact one of the UCCE Agronomists listed on this webpage.