*2.1. Species Description*

*V. dingleri* is [11] a perennial herb species that grows up to approximately 60 cm. The stem is erect, slender, with parallel lines or grooves. The inferior part of the stem is greyish with stellate hairs, without glands, and the upper part is smooth and without hairs. The basal leaves (rosette) are lax with stellate hairs on both sides, denser on the lower side. The shape of the basal leaves is oblong, inversely ovate, the inferior part pinnate, with a length of approximately 12 cm and a width of 3–5 cm. Basal leaves with peduncles have a length of about 3 cm and a width of about 0.5 cm. Leaves' base is flat above and convex underneath. The wings on either side of the petiole inflorescence with simple or complex panicles more or less branched, open, erect and slender, 15–30 cm long, racemose. The flowers with peduncles can be solitary, 5–7 mm long, with a little bract that appears ovate-lanceolate, acute, without hairs. The calyx is more or less conic, with a length of 3–3.5 mm and a diameter of 2.5 mm, without hairs. The calyx lobes are linear-lanceolate, acute or sub-obtuse, with a length of 2.8–3 mm and a width of 0.8–1 mm, and present margins with minutely sparse glands, with obscure three-nerved and middle-nerve keel. The corolla is yellow, about 1.5–2 cm in diameter, without hairs on both sides; the corolla tube is about 3 mm long, with rounded inversely ovate lobes. The anthers are reniform, about 1.5–1.8 mm long, the stamens filament are about 3 mm long and 2.5 mm wide. The filaments are pale-yellow (sometimes whitish), dry, and dense with clavate apex. The bauds are globose with slender, dense, stellate greyish hairs. The style apex is flattened-clavate, about 8–9 mm long. The mature capsule is ovate-globose, about 7 mm long and 5.5 mm in diameter, partly glabrous, the style is persistent. The seeds are ovate-turbinate, intensively warty and gibbous, about 1–1.2 mm long and

0.6–0.7 mm wide [11]. Flowering lasts from late May to the end of June, and fruiting from mid-July to early September.

## *2.2. Species Distribution*

Based on the available information, a survey for species appearance was carried out in 2016 and 2017 in a wide area around the three locations recorded in the GBIF (2018) in northeastern Greece. The surveys were carried out on foot by two people in the locations mentioned in the GBIF and in nearby similar areas on the basis of site-related similarities. Special emphasis was given to locations with topographical and ecological characteristics similar to those of the suggested locations for the species, such as altitude, slope aspect, topography, geological and edaphic characteristics, vegetation type and land uses. It should be noted that the reported coordinates of the first species record (in Chrysoupolis) correspond to a flat, agricultural land, 20 m above sea level. Probably, some correction of the coordinates is possible taking into consideration the year of the report (1936).

#### *2.3. Estimation of Environmental Drivers Limiting the Species Distribution*

In the locations where the species was found, we recorded the geographical coordination by GPS, the altitude, the topographical characteristics, the slope aspect and inclination, and the soil depth. For local climate estimation, the meteorological data of the nearest meteorological station of Kavala were used, which is at a distance of 5 km from the western area of species distribution and 25 km from the eastern limits, at similar altitude, latitude, and distance from the sea. To estimate the specific soil conditions of the species habitat, a soil sampling was carried out in 2017. Four surface soil samplings were made from four locations, with three replicate samples per location. The soil properties were measured (texture, pH—using the 1:5 (weight/volume) method—total nitrogen, phosphorus, and potassium concentrations) by standard methods for soil analysis [15,16]. In addition, because of the extensive rock presence on the soil surface, a visual estimation of rock percentage covering the sampling area was carried out.

#### *2.4. Community Characteristics and Estimation of Possible Biotic Interactions*

To gain a good understanding of the vegetation existing in the species habitat, we used the sampling method of Braun–Blanquet and a specific, modified abundance/dominant scale [17]. Thus, a full record of phytosociological data was made in five plots, sized 100 m<sup>2</sup> [18], in the summers of 2016 and 2018. Floristic elements where collected in the field, while plant taxa identification was made at species level in the laboratory. The plant species found were analyzed according to their functional attitudes, life form, and any possible interactions with *V. dingleri*.

#### *2.5. Fruit and Seed Collection and Laboratory Analysis*

We measured the percentage of individuals bearing fruits among 100 randomly selected individual plants and measured the number of fruits for 30 individuals. In early July 2016, only the minimum number of most of the mature fruits was collected (approximately 60–70 fruits from the most productive plants), since in the case of seeds and fruits belonging to rare and protected species, their collection and use in experiments should be limited to minimum [19]. The collected fruits were put in sealed plastic bags, transported to the laboratory (Aristotle University of Thessaloniki) on the same day of collection, and put in a refrigerator. The size (diameter) of all collected fruits was measured using a digital caliper with accuracy of 0.1 mm. Then, they were classified in three size classes, according to their diameter: a large class, with diameter (d) over 3.5 mm, a medium class, with 3.5 mm > d > 3.0 mm, and a small class, with d < 3.0 mm. Afterwards, the seeds were carefully extracted from the fruits and separated from the peel, and the amount of seeds per fruit was measured. The seeds of each fruit were then set separately in small paper bags.

The morphological characteristics of seeds (length, weight, and water content) were determined in a sampling of 15 seeds of five randomly selected fruits per fruit class (225 seeds in total). The seed number per capsule was counted using a stereomicroscope (magnification range 6.7–45×). The floating method was used for seed purity estimation; only high-quality mature seeds were selected for the test. Then, the length and the fresh weight of fully developed seeds were measured in each fruit class. The seed water content was determined following standard laboratory procedures. The seeds were gravimetrically dried at 72 ± 2 ◦C for 72 h [20], then the final seed water content was calculated on a dry mass basis (%). All seeds were stored in the refrigerator at 4 ◦C up to the initiation of the germination examinations (three months later).

#### *2.6. Seed Germination under Controlled and Ambient Environmental Conditions*

Before assessments, the seeds were surface-sterilized using 0.85% sodium hypochlorite for 1 min, after which they were washed with distilled water. Four replicates of 25 seeds for each of the three fruit classes were placed in glass Petri dishes (9 cm diameter) containing a layer of filter Whatman paper wetted with distilled water. Parafilm M® was used for wrapping the Petri dishes to restrict any moisture loss, while distilled water was added as needed to provide seeds with an adequate moisture level. The Petri dishes were placed in a plant growth chamber at a constant temperature of 20 ◦C. This temperature was selected on the basis of the existing data for other species of the genus *Verbascum* [21,22]. The fruit size effect on seed germinability was studied by testing the seeds of four fruits per fruit class (1st, 2nd, and 3rd). Seed germination was checked every two days; water was added as needed during the period of the germination test. The criterion for establishing germination was the emergence of a radicle with length of approximately 2 mm [23]. The experiment was terminated when no seeds germinated for one week. The cumulative germination percentage was evaluated every two days, and the final germination after 28 days. The germination percentage was calculated as the average of four replicates of 25 seeds according to Equation (1), and the mean germination time (MGT) was calculated according to Equation (2) [24,25].

$$\text{GP} \left( \% \right) = \left( \text{number of permuted seeds} \text{\#total seeds per sample} \right) \times 100 \tag{1}$$

$$MGT = \Sigma(t.n)\Sigma n \tag{2}$$

where *t* is the time (days) from the beginning of the test to the end of the assessment, and *n* is the number of germinated seeds on day *t*.

#### *2.7. Seed Germination and Seedling Emergence at Nursery Conditions*

Nine fruits were randomly selected (three from each size class), and a random sample of 15 seeds was taken from each of them (in total, 135 seeds). The seeds were planted in plastic pots (Quick pots of 24 cavities with cell volume of 330 cm<sup>3</sup> and depth of 16 cm) in an open-air nursery (research forest nursery of the Laboratory of Silviculture of Aristotle University of Thessaloniki), under relatively similar climatic conditions (similar type of climate, same latitude, closed to the sea), in March 2017. The pots were filled with a common growing medium consisting of peat/perlite in a ratio of 3:1 v/v. The position of the pots in the nursery was changed periodically. All pots were watered to field capacity. After one month, the number of fully developed seedlings (shoot with leaves) per fruit was recorded.

## *2.8. Statistical Analysis*

Statistical analysis of the data was performed using the SPSS software (version 23.0, SPSS Inc., Chicago, IL, USA). Before the analysis, the percentage values of seed germination and seedling emergence were arcsine-transformed to cover the normality and homogeneity assumptions. Seed morphological data as well as the transformed values of seed germination and seedling emergence were subjected to one-way analysis of variance to detect any differences between fruit classes. Comparison of the means followed the least significant differences (LSD) criterion (0.05 level of probability).
