Defining Associations Between Berry Features of Wild Red Currant Accessions Utilizing Various Statistical Methods

Creative Commons License

Akin M., Eyduran S. P., Gazioğlu Şensoy R. İ., Eyduran E.

ERWERBS-OBSTBAU, vol.64, no.3, pp.377-386, 2022 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 64 Issue: 3
  • Publication Date: 2022
  • Doi Number: 10.1007/s10341-022-00660-3
  • Journal Name: ERWERBS-OBSTBAU
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Agricultural & Environmental Science Database, CAB Abstracts, Environment Index, Food Science & Technology Abstracts
  • Page Numbers: pp.377-386
  • Keywords: Currant, Berry characteristics, Fuzzy clustering, Hierarchical cluster analysis, Factor analysis, FACTOR-ANALYSIS SCORES, MORUS-NIGRA L.
  • Van Yüzüncü Yıl University Affiliated: Yes


This research was performed to define genetic diversity of wild red currant accessions native to Eastern Anatolia region of Turkey by revealing associations between physicochemical berry characteristics through various statistical methods including Explanatory factor analysis, hierarchical cluster analysis (single linkage-Euclidian distance) and fuzzy clustering (Manhattan distance). The factor analysis explained 89% of the data variability on the tested berry features by four factors. The first factor was called berry color and showed positive loadings on A (0.947), B (0.925) and negative loading on L (- 0.909). The second factor was named organoleptic which had positive loadings on aroma (0.993) and taste (0.993). The third factor was called pomology and pH and demonstrated positive loadings on fruit weight (0.903) and fruit length (0.824), but negative loading on pH (- 0.583). The fourth factor was named soluble solid content and exhibited positive loading of 0.928. The hierarchical cluster analyses resulted with seven clusters showing 83.89 (%), 83.33 (%), 80.71 (%), 88.21 (%), 91.28 (%), 95.12 (%) and 90.37 (%) similarity for the first, second, third, fourth, fifth, sixth and seventh clusters, respectively. The largest values of average silhouette and Dunn's partition coefficient, as well as the smallest value of the normalized partition coefficient of fuzzy clustering analysis resulted in five clusters. Furthermore, a hybrid approach combining fuzzy clustering and decision tree algorithms was established to better characterize the phenotypical profile of red currants. We can conclude that the statistical methods utilized in this research could be a useful tool in revealing phenotypic similarities and differences among red currant accessions, and the knowledge on berry feature associations could be helpful in plant breeding programs.