Lorenzo Della Maggiora

The Pistoia Mountains exhibit a great variety of flora, particularly rich in aromatic plants, such as juniper, mint, savory, helichrysum, and thyme. Thyme is especially notable for its essential oil, typically displaying high thymol and carvacrol content. While the chemotype of thyme determined by its specific terpene composition is genetically controlled, environmental factors, plant age, and seasonality can influence terpene production. This article investigates the morpho-chemical variability of wild thyme plants collected from two different regions of the Pistoia Mountains, identifying five distinct chemotypes. The gas chromatography–mass spectrometry (GC-MS) technique was used to characterize the chemical profiles and determine the seasonal variation in terpene production, identifying spring and summer as the balsamic period, the optimal time for essential oil collection. Furthermore, high-value thyme clones were preserved through in vitro micropropagation, ensuring chemotype stability. These findings contribute to a deeper understanding of wild thyme biodiversity and provide a foundation for practical applications, including the development of value-added products like herb-infused cheeses, plant and animal disease treatments, and integrated pest management strategies in agricultural systems.

In recent times, salinity has emerged as a significant abiotic stress affecting plant survival in an ecosystem. In the experimental setup, response to salinity was investigated in Populus alba L. ‘Villafranca’ clone transgenic plants (line 16, 22 and 24) over-expressing an aquaporin – aqua1 – and compared to P. alba wild-type (Wt). Plants were grown in vivo under both control conditions (0 mM NaCl) and salinity (100 mM NaCl) for the duration of 28 days. This work presents a pilot study to investigate the performance of machine learning based algorithms to classify poplar leaves exposed to salt stress treatment, with the aim of performing the task in automated manner. The data set used in this work contains images acquired through a normal RGB camera. Segmentation and normalization based preprocessing were applied on the complete data set. Convolutional Neural Networks (CNNs) based architectures were employed for the extraction of prominent features from the images. Random Forest (RF), Support Vector Machine (SVM) and Logistic Regression (LR) classified the extracted feature into the group of plants treated with 100 mM NaCl or otherwise. The primary evaluation index, accuracy indicates that the classification with SVM and RF achieved highest accuracy 75% and 71% to detect leaves of 100 mM NaCl treated plants.