The Mediterranean islands represent a center of vascular plant diversity featuring a high rate of endemic richness. Such richness is highly threatened, however, with many plants facing the risk of extinction and in need of urgent protection measures. The CARE-MEDIFLORA project promoted the use of ex situ collections to experiment with in situ active actions for threatened plants. Based on common criteria, a priority list of target plant species was elaborated, and germplasm conservation, curation and storage in seed banks was carried out. Accessions were duplicated in the seed banks of the partners or other institutions. Germination experiments were carried out on a selected group of threatened species. A total of 740 accessions from 429 vascular plants were stored in seed banks, and 410 seed germination experiments for 283 plants species were completed; a total of 63 in situ conservation actions were implemented, adopting different methodological protocols. For each conservation program, a specific monitoring protocol was implemented in collaboration with local and regional authorities. This project represents the first attempt to develop common strategies and an opportunity to join methods and methodologies focused on the conservation of threatened plants in unique natural laboratories such as the Mediterranean islands.

Pine honey is a honeydew honey produced in the East Mediterranean region (Greece and Turkey) from the secretions of the plant sucking insect Marchalina hellenica (Gennadius) (Coccoidea: Marchalini-dae) feeding on living parts of Pinus species. Nowadays, honeydew honey has attracted great attention due to its biological activities. The aim of this study was to study unifloral pine honey samples produced in Greece regarding their physicochemical parameters and antioxidant and antibacterial activity against five nosocomial and foodborne pathogens. These honeys showed physicochemical and microscopic characteristics within the legal limits, except for diastase activity, a parameter known to be highly variable, depending on various factors. Substantially higher levels of H₂O₂ were estimated compared to other types of honeydew honey, whereas protein content was similar. The total phenolic content was 451.38 ± 120.38 mg GAE/kg and antiradical activity ranged from 42.43 to 79.33%, while FRAP values (1.87 to 9.43 mmol Fe⁺²/kg) were in general higher than those reported in the literature. Various correlations could be identified among these parameters. This is the first attempt to investigate in depth the antibacterial activity of pine honey from Greece and correlate it with honey quality parameters. All tested honeys exerted variable but significant antibacterial activity, expressed as MIC and MBC values, comparable or even superior to manuka honey for some tested samples. Although honey antibacterial activity is mainly attributed to hydrogen peroxide and proteins in some cases (demonstrated by elevated MICs after catalase and Proteinase K treatment, respectively), no strong correlation between the antibacterial activity and hydrogen peroxide concentration or total protein content was demonstrated in this study. However, there was a statistically significant correlation of moisture, antioxidant and antibacterial activity against Klebsiella pneuomoniae, as well as antioxidant and antibacterial activity against Salmonella ser. Typhimurium. Interestingly, a statistically significant negative correlation has been observed between diastase activity and Staphylococcus aureus antibacterial activity. Overall, our data indicate multiple mechanisms of antibacterial activity exerted by pine honey.

Pollen identification is an important task for the botanical certification of honey. It is performed via thorough microscopic examination of the pollen present in honey; a process called melissopalynology. However, manual examination of the images is hard, time-consuming and subject to inter- and intra-observer variability. In this study, we investigated the applicability of deep learning models for the classification of pollen-grain images into 20 pollen types, based on the Cretan Pollen Dataset. In particular, we applied transfer and ensemble learning methods to achieve an accuracy of 97.5%, a sensitivity of 96.9%, a precision of 97%, an F1 score of 96.89% and an AUC of 0.9995. However, in a preliminary case study, when we applied the best-performing model on honey-based pollen-grain images, we found that it performed poorly; only 0.02 better than random guessing (i.e., an AUC of 0.52). This indicates that the model should be further fine-tuned on honey-based pollen-grain images to increase its effectiveness on such data.