Functional differentiation in pollination processes among floral traits in Serapias species (Orchidaceae)

Abstract Floral displays, influencing attractiveness to insects, increase the number of pollinator visits and the efficiency of each visit in terms of pollen exchange and thus affect the plant reproductive success. Here, we conducted an in situ manipulation experiment to investigate whether the floral modifications affect reproductive success in natural orchid populations of Serapias lingua and Serapias vomeracea. We estimated male and female reproductive success of three treatment groups, disassembly of floral tube, cutting of lip, and painting of the callus surface, in terms of pollinaria removed/deposited and fruit production. Results revealed that phenotypic modification had opposite effects on reproductive success of two examine species. Indeed, reproductive success was significantly increased by the detached of the petals and sepals, and decreased, due to callus painting and lip removal, in S. lingua. On the contrary, unmanipulated plants of S. vomeracea showed significantly higher value of pollinaria removed and deposited and fruit set than manipulated ones. The differences between S. lingua and S. vomeracea agree to the different pollination strategy of examined species. S. vomeracea shows shelter imitation strategy, and thus, the disassembly of tunnel‐like corolla does not allow the insects to use the flower as a refuge, while S. lingua is a sexually deceptive orchid and therefore the opening of the flower made more visible callus (visible at a greater distance) increasing the pollinators attraction. This study provides evidence that pollinators were largely sensitive to the experimental modification of the flower phenotype, which is consistent with the presence of significant selection on individual floral characters. Our experimental investigations of the effects of variation in display on pollinator visitation provide insights into the evolution of floral morphology in orchid with shelter imitation strategy.

deceptive genus Ophrys that shows a close morphological and olfactory resemblance between mimic and model species (Tang, Ou, Luo, Zhuang, & Liu, 2014). Sexually deceptive orchids mimic signals emitted by female insects in order to attract mate-searching males. The olfactory compounds produced by the labellum of the orchids act as long-range attractants, guiding males to the proximity of flowers. At close range, also in sexually deceptive orchids, floral visual signal plays a key role in pollinator attraction and thus affects reproductive success (Rakosy, Streinzer, Paulus, & Spaethe, 2012).
In the last years, the pollination strategy of Serapias genus has received more attention, pointing out that Serapias species show two different pollination strategies. Indeed, Serapias lingua seems to have evolved sexually deceive pollinators, analogous to what is observed in Ophrys (Vereecken, Dafni, & Cozzolino, 2010).
This phenomenon is also supported by the finding of large amounts of alkanes and alkenes in its floral odor extracts (Pellegrino, Luca, Bellusci, & Musacchio, 2012;Schiestl & Cozzolino, 2008), and Ceratina cucurbitina males are its main pollinators (van der Cingel, 2001;Paulus, 2014;Vereecken et al., 2010). While an unusual type of deceptive pollination mechanism has been observed in the other Serapias species called shelter imitation strategy (Jersáková et al., 2006). In these cases, the sepals, petals, and lateral lobes of the hypochile form a tunnel-like corolla, varying in diameter and depth among taxa, offering a floral tube to the insects in which they use to rest or sleep under bad or rainy weather conditions (van der Cingel, 1995).
Here, we assess the potential role of pollinator-mediated selection on floral morphology through female and male reproductive success using experimental manipulations of floral traits. To examine the functions of attractive structures that differ in shape, it is necessary to investigate stages of the pollination process in relation to each attractive structure. We focus on three floral traits, broadly related to pollinator attraction: the form of flowers, the color of callosity at the base of the hypochile, and the presence of the lip. We manipulated the form of the Serapias flowers, transforming them from tubular form to open flowers, eliminating the lip and coloring the callosity, to examine the effects of such changes in attracting pollinators, and to quantify the effects on male and female reproductive success, that is, pollen removal and fruit/seed set, respectively. Therefore, the aims of this study were to evaluate the effects of floral morphology and the role of the lip and the callus on pollination activity and to assess (1) if pollinators increase the rate at which they visit the flowers, as a result of responding to a more attractive "visual signal," or (2) pollinators decrease their visits to modified flowers which are less attractive than the natural ones. We carried  The genus Serapias L. is distributed throughout the Mediterranean region with its centre of diversity in Southern Italy and on the Greek islands (Baumann & Künkele, 1989). A more recent systematic treatment included 30 taxa (Delforge, 2006), which were characterized by a common floral morphology, the lateral petals, and hypochile (the proximal part of the lip) form a hood (tubular corolla). Recent molecular analysis strongly supports a natural split of S. vomeracea and S. lingua into two different groups strictly related to two pairs of endemic species, S. apulica and S. neglecta, S. gregaria and S. olbia, respectively (Bellusci, Pellegrino, Palermo, & Musacchio, 2008).

| Natural history
Serapias vomeracea (Burm.) Briq. is a widespread species that grows in arid meadows, abandoned agricultural soils, garigues, and bushy environments up to 1000 m asl. It has a three-lobed lip without a spur, a plain-colored epichile more than 13 mm long and two guiding swellings at the base of the lip. The chromosome number is 2n = 36 Serapias lingua (tongue orchid) is a short-lived tuberous orchid and a tetraploid species (D'Emerico et al., 2000). It has dull-colored flowers of uniform structure, with a callosity at the base of the hypochile, and the epichile (the distal part of the lip) generally inclined downwards.
Conical epidermal papillae and two types of trichome with secretory apical cells characterize the petals and lip (Barone Lumaga et al., 2012). It grows in arid meadows, abandoned agricultural soils, garigue, and bushy environments up to 1,200 m a.s.l. (Delforge, 2006).
Both species are widespread species, co-occur in the same habitat, mainly distributed in the Mediterranean-Atlantic countries (Portugal, Spain, France, Italy, Balkans, Greece), but reaching western North Africa (Morocco, Tunisia), and show high level of inbreeding depression (Bellusci, Pellegrino, & Musacchio, 2009).
The pollination biology of Serapias orchids (included S. vomeracea) in the Mediterranean is usually viewed as a generalized mimicry of nests and shelter. Indeed, a wide range of insect pollinators, both males and females, were found immobile into the tubular flowers overnight with pollinaria attached to their head (Pellegrino, Gargano, Noce, & Musacchio, 2005;Pellegrino, Noce, Bellusci, & Musacchio, 2006).
Moreover, the dark-colored flowers accumulate warmth via the sun beams at dawn, providing the pollinators with a morning dose of heat presumably sufficient for them to start foraging earlier than others kept at a lower ambient temperature (Dafni, Ivri, & Brantjes, 1981). At least one species in the genus Serapias, namely S. lingua, seems to have evolved to sexually deceive pollinators, analogous to what is observed in Ophrys orchids (Jersáková et al., 2006). Moreover, preliminary observations indicate that C. cucurbitina males are the main pollinators (van der Cingel, 2001;Vereecken et al., 2010Vereecken et al., , 2012.

| Manipulative experiments
To ensure that potential selection on flower traits occurred in relation to floral function, we modified traits beyond the range naturally Each species were analyzed separately using two-way ANOVA to compare treatments (the three treatments of manipulation vs. the unmanipulated controls), to examine the percentage of flower manipulated (100%, 50% and 10% vs. the natural condition) and their interaction on proportion of pollinaria removed, proportion of flowers receiving pollinaria, and fruit formation. All proportional data were arcsine square-root transformed prior to analyses. When the F test was significant, means were compared using the Tukey test at 5% error probability.

| RESULTS
Reproductive success in natural conditions of our study species ranged from 29% to 38.3% without differences between examined orchids. In particular, in S. vomeracea 38.3% of the flowers had pollinaria removed, 38.3% were pollinated and 32.3% produced fruits; in S. lingua, 35.3% of the flowers showed pollinaria removed, 33% received pollinaria, and 29% produced fruits.
Differences among the manipulation groups were highly significant in both species (Tables 1 and 2). In particular, intact flowers of S. vomeracea had significantly higher value of pollinaria removed and deposited and fruit set than those with manipulated flowers, while S. lingua flowers with the petals and sepals detached and the tube shape opened showed significantly higher value of pollinaria removed and deposited and fruit set than intact flowers (Figure 3).
T A B L E 1 ANOVA results of the effects of treatments (OPN, LAB, and CAL), the percentage of flower manipulated (100%, 50%, and 10%), and their interaction on proportion of pollinia removal, proportion of flowers receiving pollinia, and fruit set in Serapias vomeracea

| DISCUSSION
The percentage of fruits produced by the natural populations of examined species (29%-32.3%) is within the ranges reported for nectarless orchids of temperate zones (Neiland & Wilcock, 1998) and is consistent (just a little higher) with records for previously papers on Serapias populations (Bellusci, Pellegrino, Palermo, & Musacchio, 2010;Pellegrino, Musacchio, Noce, Palermo, & Widmer, 2005). Instead, S. lingua is a sexually deceptive orchid, and therefore, the opening of the flower made more visible callus (visible at a greater distance) and increased the intensity of scents. Insect males, searching nesting sites or mates, orientate themselves by odor (Ayasse et al., 2000;Schiestl, 2010), but at short distances, when objects subtend a F I G U R E 3 Variation of reproductive success among manipulation, OPN (the petals and sepals have been detached, and the tube shape was opened), LAB (lip was completely removed), and CAL (the flower was open, and the callus surface was painted) and NAT (natural condition) in Serapias lingua F I G U R E 4 Effects of three manipulation experiments on Serapias lingua. OPN (the petals and sepals have been detached, and the tube shape was opened), LAB (lip was completely removed) and CAL (the flower was open, and the callus surface was painted) performed on all flowers (100%), on half flowers (50%), and on 10% of flowers of an inflorescence certain visual angle, optical features can be perceived and act as shortrange stimuli (Streinzer, Paulus, & Spaethe, 2009). Thus, the increased male and female reproductive success associated with two of the three trait manipulations can be readily explained by increased attractiveness to pollinators. Pollinators first approach the S. lingua flowers guided by the pheromone produced by the flowers, but when close enough, they prefer flowers with open perianth and with the callus exposed.
In the first case, the results highlight that also for an orchid with ref- uge strategy (such as S. vomeracea), visual signals play a key role in pollinator attraction as already demonstrated in rewarding (Duffy & Stout,