**1. Introduction**

Orchidaceae are one of the largest families of vascular plants [1]; since Darwin [2], they have attracted the interests of a plethora of naturalists for the amazing floral variations and the complex pollination mechanisms which they evolved [3]. Approximately one-third of Orchidaceae are believed to deceive insect pollinators [4,5]; among the mechanisms of deception, generalized food deception is one of the most common mechanisms developed by orchids for efficient pollen exportation [5,6]. These species can exploit the existing plant–pollinator relationships and achieve pollination through deception in the absence of floral rewards for pollinators. The similarity with rewarding plants determines their reproductive success; therefore, this pollination syndrome can be considered a generalized form of Batesian mimicry [7]. In order to deceive pollinators, these orchids exploit general floral signals typical for rewarding plant species, including flower color and scent [8]. However, they generally do not resemble any specific rewarding flower and they are visited by casual pollinators or exploratory pollinators [9]. For an example of floral mimicry, see [10] and references therein. It has been shown that orchids related to generalized food

**Citation:** Romano, V.A.; Rosati, L.; Fascetti, S.; Cittadini, A.M.R.; Racioppi, R.; Lorenz, R.; D'Auria, M. Spatial and Temporal Variability of the Floral Scent Emitted by *Barlia robertiana* (Loisel.) Greuter, a Mediterranean Food-Deceptive Orchid. *Compounds* **2022**, *2*, 37–53. https://doi.org/10.3390/ compounds2010004

Academic Editor: Juan Mejuto

Received: 13 November 2021 Accepted: 4 January 2022 Published: 27 January 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

133

deception exploit bees and bumblebees that have just emerged after the winter season, blooming in early spring [11]. Moreover, reward-less species undoubtedly benefit from the simultaneous flowering of nectariferous species present in the same habitat that increase the possibility of being visited by local pollinators [12].

Floral scent, together with size, shape and color, act as signals attracting pollinators [13], and adaptations to specific pollinators are considered an important driver of evolution in angiosperms.

*Barlia robertiana* (Loisel.) Greuter (Orchidaceae) has non-rewarding flowers, and it is obligatorily insect-pollinated. It is a Mediterranean species [14] typical of several habitats such as clearings in scrublands and thermophilus woods, dry grasslands, roads edge, usually on bases-rich soils [15]. Recent extensions north due to climate change have been observed in western Switzerland [16] and southwestern Germany [17]. Its geographical distribution in Italy encompasses all the regions, but is lacking in some of the Alpine sectors [18]. It is particularly widespread in the Basilicata region (Southern Italy), from the coast to the mountains of the Apennines up to approximately 1000 m a.s.l. (above see level).

According to a recent molecular genetic study [19], the genus *Barlia* should be transferred to the genus *Himantoglossum*; however, a general consensus about this taxonomical rearrangemen<sup>t</sup> has not been reached by taxonomists; in this article, we refer to the nomenclature of the most recent checklist of the Italian vascular flora [18].

The plants are robust (up to 80–110 cm high) and early flowering, from December to April. The inflorescence is sub-cylindrical, dense and multi-flowered; it is up to 40 cm high and can develop up to 70 flowers; the color of flowers varies from red-violet to olive-green or brown-red; inside, it is covered with purplish spots [14]. The lip borders are crenate, with a papillose epidermis; the spur is conical, shorter than the ovary, turned downwards, and it does not produce nectar. The flowers give off a delicate, persistent, and easily perceptible smell. Structural particularities of epigeous and hypogeous plant organs of *B. robertiana,* in comparison with *H. hircinum*, have been interpreted as morphological adaptations to different edaphic and environmental conditions [20]. Hydroalcoholic flower extracts of *B. robertiana* revealed the presence of phenols, flavonoids, and proanthocyanidins [21]. Chromosomes (2*n* = 32), karyotypes, and the localization of ribosomal genes have been studied [22]. Unlike other Mediterranean orchids, such as the genus *Ophrys*, which are pollinated by specialized bees [3], several different groups of insects have been identified to pollinate *B. robertiana*, such as Apoidea (Hymenoptera) and Cetoniidae (Coleoptera) [23,24]. A recent study on the fruit set of *B. robertiana*, performed on the island of Mallorca (Spain), confirmed the importance of allogamy for its reproduction success [25].

Although the scent of orchids has frequently been analyzed in evolutionary studies or to identify potential chemical fragrances, knowledge about most of the species are still incomplete; moreover, studies of floral scent variations at population level including more than just a few individuals are particularly scarce. In particular, knowledge about the spatial and temporal variability of the floral scents emitted by flowers is almost completely absent.

In this study, we aim to describe the spatial and temporal variability of the spectrum of volatile compounds emitted by flowers of *B. robertiana,* testing the ability of this species to adapt to different environmental conditions to be attractive to different potential pollinators.

*B. robertiana*, with respect to other groups of orchids, is not plagued by taxonomic problems at species level that could determine some confounding effects due to identification discrepancies by the botanists. Moreover, it has large populations in southern Italy, and it has a broad range, spreading along a large climatic gradient, at regional level. Thus, it represents an ideal case study to explore the temporal, ecological, and spatial variability in emitted floral volatile compounds (VOCs). In particular, in this study, we aim to: (1) assess whether the chemical composition is variable between plants collected at different sites; (2) evaluate whether chemical composition of the scent is related to geographical or environmental variables; (3) assess whether there are temporal differences in VOC compositions during the flowering phase, comparing floral volatile compounds emitted at the beginning

and at the end of the flowering; and (4) assess whether there is a between-year variability in VOC compositions emitted by the same plant.

#### **2. Experimental Section**

To assess spatial variability, fourteen *B. robertiana* plants were collected from different populations in the Basilicata region (Southern Italy) from the coast to the inner mountain area (Figure 1) within altitude ranges from 10 m a.s.l. to 727 m a.s.l. Plants were collected in 2017 and cultivated into pots at the campus of the University of Basilicata (Potenza, Italy). In the following year (2018), inflorescences with fully opened flowers were incapsulated in a of 6.5 glass bell (Figure 2). Sampling was performed under light conditions, in an air-conditioned room (21 ± 1 ◦C) to guarantee a stable temperature. VOC sampling followed the protocol we used in a recent study on the floral volatiles of the genus *Gymnospermium* [26].

**Figure 1.** Study area and sampling point locations in the Basilicata region (southern Italy). Labels of the samples correspond with the municipality names where plants were collected.

Analysis of VOCs was performed using HS-SPME with a DVB/CARB/PDMS fiber. A preliminary set of analysis was conducted to optimize the sampling time: these analyses were performed at three different adsorption times of the fiber (5–24–72 h). The highest number of identified compounds (20) was detected when the SPME fiber was exposed for 24 h (Appendix A, Table A1); therefore, we used this interval of fiber exposition in this study. The fiber was exposed to the headspace and then withdrawn into the needle and transferred to a GC/MS system. A 50/30 μm DVB/CAR/PDMS module (57328-U, Supelco, Milan, Italy) was employed to determine VOCs. Analyses were accomplished with an HP 6890 Plus gas chromatograph (Agilent) equipped with a Phenomenex Zebron ZB-5 MS capillary column (30 m × 0.25 mm i.d. (inner diameter) × 0.25 μm FT) (Agilent, Milan, Italy). An HP 5973 mass selective detector (Agilent) was utilized with helium at 0.8 mL/min as the carrier gas. A splitless injector was maintained at 250 ◦C and the detector at 230 ◦C. The oven was maintained at 40 ◦C for 2 min, then gradually warmed, 8 ◦C/min, up to 250 ◦C and held for 10 min (Figure 3). Tentatively identification of aroma components was based on mass spectra and NIST 11 library comparison. A single VOC peak was considered as identified when its experimental spectrum matched with a score over 90% with ones present in the library and if the retention time was in agreemen<sup>t</sup> with the reported retention index. Retention indices were calculated using standard *n*-alkane solution (49452-U, C7-C40 saturated alkanes standard, Sigma-Aldrich, Milan, Italy).

**Figure 2.** A *Barlia robertiana* plant from the site of S. Arcangelo (Basilicata, Italy) in the experimental conditions for floral VOC sampling using a 50/30 μm DVB/CAR/PDMS fiber (photograph: V.A. Romano).

A multivariate analysis of VOC compositions was performed through non-metric multidimensional scaling (NMDS) as an ordination technique and hierarchical cluster analysis (HCA). Bray–Curtis dissimilarity was used for both NMDS and HCA using the relative abundance of each compounds as sample variable. For HCA, we used the unweighted pair group method with arithmetic mean (UPGMA) technique as an agglomerative method. Variables with Spearman's rho correlation coefficient > 0.65 with ordination axes were superimposed in the scatter diagram.

Concerning spatial analyses, we tested whether geographic distance and environmental variables influence the VOC compositions of samples. Site-specific soil data were not available; therefore, we approximated differences in the environmental niche using slope, aspect, altitude, and bioclimatic data. For each sample point, the most relevant phytoclimatic indices [27,28] were extracted from the high-resolution raster dataset developed to realize the bioclimate map of Italy [29]. These indices were: yearly positive temperatures (Tp = sum of the monthly mean temperatures of months with average temperatures > 0 ◦C); annual positive precipitation (Pp = total average precipitation of months with average temperature > 0 ◦C); thermicity index (T + m + M); annual ombrothermic index = (Pp/Tp); continentality index (Tmax − Tmin); ombrothermic index of the warmest summer bimester (Ios2 = (Pp2/Tp2). According to the approach we used for population genetics of *Centaurea filiformis* [30], the correlations between VOC compositions and both geographic and environmental distances were assessed using Mantel tests implemented in the software PAST. Geographic distances were log-transformed and

environmental distances were obtained by means of analyzing the Euclidean distance after data standardization.

**Figure 3.** Gas chromatographic results of an SPME analysis of the VOCs of *Barlia robertiana*. 1: tetradecane; 2: caryophyllene; 3: sesquiphellandrene; 4: β-himalachene; 5: 1-(1,5-dimethyl-4- hexenyl)-4-methylbenzene; 6: pentadecane; 7: β-bisabolene; 8: β-sesquiphellandrene; 9: hexadecane; 10: phthalate; 11: *trans*-farnesol; 12: 2,3-Dihydrofarnesol; 13: 2,6-diidopropylnaphthalene; 14: tetradecanoic acid; 15: octadecane; 16: pentadecanal; 17: phthalate; 18: nonadecane; 19: phthalate; 20: eicosane; 21: heneicosane; 22: docosane; 23: phthalate; 24: phthalate; 25: phthalate.

To assess temporal variability of flower scent in *B. robertiana*, two individuals collected from the S. Arcangelo population were sampled twice during 2019, following the same protocol described above. The first sampling was performed in the first part of the flowering phase, approximately when 50% of the flowers were open; the second sampling at the end of the flowering, when all the flowers were open. In addition, three individuals from Potenza, Calciano and S. Arcangelo were sampled in two consecutive years (2018 and 2019), in the same experimental conditions described above, to assess the between-year stability of floral emissions. Similarity between samples has been measured through the Bray–Curtis index.
