Determination of duplicates of accessions in a germplasm collection of flax/linseed by means of digital image analysis

The use of digital image analysis (DIA) for characterization of varieties is much quicker and easier than conventional morphometric methods. This non-destructive, highly effective technique employs an inexpensive imaging system, which allows image acquisition, image processing and feature extraction. In this study digital image analysis was used for the detection of assumed duplicates in the germplasm of a flax/linseed collection, based on petal measurements. A group of 131 different flax/linseed varieties was used in tests for defining key features. The aim of the investigation was to determine if petal size could be used to distinguish compared accessions (doubles, triples, quadruples). The linear regression model used for feature assessment enabled successful identification of the dissimilarity or similarity among tested accessions. The position and distance of accessions in groups of clusters were obtained by hierarchical cluster analysis. Duplicates in the germplasm of the flax collection were confirmed by non-significant statistical differences between tested accessions. Based on the passport descriptor ACCENAME analysis it was revealed that out of 97 evaluated couples 81(83.5%) showed a significant difference in petal size, resulting in the conclusion that they are not real duplicates. Only 10 couples (10.3%) showed non-significant differences in petal size and shape could be considered as real duplicates. Six couples (6.2%) remained undetermined, as it was not possible to make a clear distinction using the above mentioned methods, due to the influence of year or other external factors. Image analysis followed by hierarchical cluster analysis was recommended as one of the methods useful for the characterization and rationalization of a germplasm collection. (C) 2011 Elsevier B.V. All rights reserved.
Industrial Crops and Products. March 2012, Volume 36, Issue 1, Pages 177–187
doi: 10.1016/j.indcrop.2011.09.004
Tvůrci externí: 
David Hampel

Přečtěte si více ...