Taxonomic discrimination of the Paeonia mascula group in the Tyrrhenian Islands by seed image analysis

The taxonomy of the Paeonia mascula group in the Tyrrhenian Islands has always been controversial, especially in Sardinia and Corsica. In this paper, 124 morpho-colorimetric quantitative and qualitative features of seeds of the Paeonia mascula group from the Balearic Islands, Corsica, Sicily, and Sardinia were measured by an image analysis system to evaluate whether differences in seed morphology were present and if the results support the current taxonomic treatment. Digital images of 41 accessions for a total of 3,317 seeds were acquired using a flatbed scanner, and morpho-colorimetric seed features were analysed using a plugin developed for the open source software ImageJ. The data were statistically processed by applying the stepwise Linear Discriminant Analysis (LDA). The results showed that the seeds of the studied taxa were distinguishable with a high percentage of classification and highlighted that three different taxa are identifiable in Corsica (P. corsica, P. mascula subsp. mascula, and P. morisii), while only one (P. cambessedesii) is identifiable in the Balearic Islands. In the same way, the morpho-colorimetric analysis identified two well-defined taxa in Sicily (P. mascula subsp. mascula and P. mascula subsp. russoi). On the other hand, the discriminant analysis suggested that only P. corsica/morisii and P. sandrae were present in Sardinia, while differences were highlighted between Sardinian and Corsican peonies. The present study represents the first attempt to investigate the taxonomy of the P. mascula group, which makes in-depth usage of the seed features as diagnostic characters. The results obtained in this work enable the usefulness of image analysis techniques in taxonomic investigations to be suggested in order to contribute to the future systematic evaluation.