Phenotypic markers for selection of winter hardy garden chrysanthemum (Dendranthema x grandiflora Tzvelv.) genotypes

Consumer interest in winter hardy herbaceous perennials has precipitated a re-prioritization of breeding program objectives for garden chrysanthemums. The objective of this study was to determine which phenotypic marker(s) in garden chrysanthemum crowns are correlated with winter hardiness for use as non-destructive, inexpensive selection tools in breeding programs. An experiment was conducted over 2 years (fall 1999, 2000) using plants grown in field trials at 45degreesN latitude (-34.5 degreesC, minimum temperature). Germplasm for fall 1999, included n = 5 numbered genotypes, n = 8 cultivars from the University of Minnesota breeding program, and n = 2 cultivars from Yoder and GroLink. In fall 2000 studies, germplasm consisted of n = 5 cultivars (University of Minnesota) and n = 10 cultivars (Yoder, GroLink). Four genotypes were included in both years. Ramets (n = 24) of each genotype were planted in May of each year. First-year, field-grown crowns (n = 3 replications/genotype) were dug each fall. Winter hardiness classifications were assigned to genotypes based on winter survival data for clones collected over the 2 years of the study. Data was collected on the number of emergent rhizomes, number of non-emergent rhizomes, proportion of total rhizomes that were emergent, fresh/dry weights, and measurements (length, diameter, % branching) of the five most basipetal rhizomes. Discriminant analysis was conducted on the pooled data for the 2 years. The data were divided into training and evaluation sets for analysis. The best discriminant function model included the variables as the number of emergent and non-emergent rhizornes, and average rhizome length. The success rate of the discriminant analysis was 88.9% for both the training and validation sets. Emergent rhizome number would be the easiest and only non-destructive trait for field selection. A second discriminant analysis was conducted using only number of emergent rhizomes. The one-variable model resulted in a success rate of 93.3% for the training set and 86.7% for the evaluation set. A garden chrysanthemum breeding program could collect data on emergent rhizome number in the fall of the first year's growth to determine winter hardiness with 86.7% accuracy, confirm this with % winter survival (in test winters), and/or laboratory freezing tests. (C) 2004 Elsevier B.V. All rights reserved.