Phenotypic identification of plum varieties (Prunus domestica L.) by endocarps morpho-colorimetric and textural descriptors

- P. domestica endocarps biometric features were measured by image analysis techniques.
- A total of 134 size, shape, color and texture descriptors were measured on each endocarp.
- The data were statistically processed applying the stepwise Linear Discriminant Analysis.
- Hypothetical synonymy groups were identified.
- This is the first computer vision application on plum biodiversity study.

The identification of plum varieties is generally done on the base of distinctive plant traits such as shape, size, and fruit drupe color identified during the variety registration, following official descriptors. In this paper, image analysis techniques were applied to study endocarps variability of 23 Prunus domestica cultivars from Sardinia. Digital images were acquired and analysed using a macro specifically developed to measure morpho-colorimetric endocarps features. The data were later statistically processed applying the stepwise Linear Discriminant Analysis (LDA) to implement a statistical classifier able to classify each variety and identify plausible synonymy groups.The present study represent the first attempt to investigate the morphology and morphometry of plum endocarps in order to characterize the whole Sardinian plum agrobiodiversity. It is also the evidence of the usefulness of image analysis techniques in taxonomic investigations too, as well as for the conservation and enhancement of traditional plums for consumer satisfaction.