Deceptive strategy in Dactylorhiza orchids: multidirectional evolution of floral chemistry

Abstract

Background and Aims

The deception strategies of orchids remain poorly understood, especially in regard to the chemical compounds emitted from their flowers and their interaction with various taxonomic groups of pollinators. We investigated the phylogenetic relationships and compared the variation of floral chemical compounds between food-deceptive Dactylorhiza taxa (D. incarnata var. incarnata and D. incarnata var. ochroleuca, D. fuchsii and D. majalis) from populations in north-eastern Poland. We propose a model of the evolution of deception based on floral chemical signals in this genus.

Methods

A Bayesian approach based on polymorphic plastid DNA (trnL, trnF and psbC–trnK), internal transcribed spacer (ITS) sequences and flow cytometry data was applied to confirm the taxonomic status of the studied orchids. We also identified and classified the pollinators and flower visitors in each Dactylorhiza population to the taxonomic level and compared our results with literature data. The chemical composition of pentane and diethyl ether extracts from the flowers was analysed by gas chromatography–mass spectrometry. Variation of the floral chemical components was visualized by non-metric multidimensional scaling based on Bray–Curtis dissimilarity.

Key Results

The genetic distinctiveness of D. incarnata, D. fuchsii and D. majalis was confirmed. No hybrids between them were found, but the chloroplast DNA (cpDNA), ITS haplotypes and flow cytometry showed genetic similarity between D. incarnata var. incarnata and D. incarnata var. ochroleuca. We determined that Apis mellifera (Hymenoptera) was the only shared pollinator of these taxa. Strangalia attenuata and Alosterna tabacicolor (Coleoptera) and Volucella pellucens and V. bombylans (Hymenoptera) were observed pollinating D. fuchsii. Visualization of the emission rates of the 61 floral chemical compounds detected from pentane extracts (mainly hydrocarbons and aldehydes) and the 51 from diethyl extracts (with abundant groups of benzenoids and non-aromatic acids) strongly differentiated D. incarnata, D. fuchsii and D. majalis, while those of the two varieties of D. incarnata (var. incarnata and var. ochroleuca) were almost identical.

Conclusions

While the genetic data clearly supported the distinct lineages of D. incarnata, D. fuchsii and D. majalis, the patterns of emission of their flower chemical compounds were more complex within the series of shared compounds (alkanes and aldehydes) and taxon-specific compounds (benzenoids and esters). Their floral bouquet can influence the sexual, social and feeding behaviour of pollinators in different ways. We observed that the floral chemical compounds attracted both shared and species-specific pollinators to Dactylorhiza, confirming the multidirectional character of floral chemical signals in these food-deceptive taxa. Reduction of species-specific pollination levels in Dactylorhiza orchid taxa may promote hybridization between them.

INTRODUCTION

In the plant kingdom, deception is a most effective and valuable reproduction strategy (Cozzolino and Widmer, 2005; Jersáková et al., 2006; Walsh and Michaels, 2017). In the Orchidaceae, a third of all species offer no reward and attract pollinators through sexually deceptive or food-deceptive pollination strategies (Tremblay et al., 2005; Inda et al., 2012). Pollination by deception is known to be a relatively widespread strategy amongst orchids (van der Pijl and Dodson, 1966; Dressler, 1981; Ackerman, 1986; Waterman and Bidartondo, 2008). Sexually deceptive orchids differ from food-deceptive plants in that they exploit the reproductive behaviour of male insects by producing sex pheromones and by presenting flower structures that mimic the shape and colour of female insects (Peakall and Beattie, 1996). Food-deceptive orchids are more common than those which are sexually deceptive; their flowers provide food-rewarding signals but do not provide the reward (Salzmann and Schiestl, 2007; Jersáková et al., 2009; Walsh and Michaels 2017). It should be stressed that the evolution of food deception in orchids cannot be explained by a single mechanism or hypothesis, because the various selective pressures that favour this strategy are highly context dependent (Tremblay et al., 2005; Scopece et al., 2007). The ‘outcrossing hypothesis’ for the evolution and maintenance of food deception in orchid species is perhaps the most empirically supported: cheated pollinators visit several flowers among the plants in populations, promoting cross-pollination (Jersáková et al., 2006). This makes inbreeding avoidance a further potential advantage of food deception. Food-deceptive orchids are pollinator generalists (pollinated by many moderately efficient pollinators) and may attract naïve, newly emerged insects (Dafni, 1984). This would explain why many food-deceptive species flower early in the season (Internicola et al., 2008). There is also evidence that experienced insects, such as bees and bumble-bees, are deceived by the flowers of these orchids (Nilsson, 1984; van der Cingel, 2001; Cozzolino et al., 2005; Peter, 2009). Gould (1993) stated that honey-bees select scent over colour and colour over shape, while in bumble-bees floral scents act as a back up to floral colour signals in low-light conditions (Lawson et al., 2017). This demonstrates how the different floral traits play crucial roles in deception.

The flower bouquet is one such trait, which varies significantly between food-deceptive orchids even within the same genus and which can influence pollinator types and their behaviours in multiple ways (Raguso, 2003; Salzmann and Schiestl, 2007). Recent studies have found that the fragrances of many orchids with food-fraud systems consist of benzenoids (phenolic compounds and phenylpropanoids) and terpenoids, as well as alkanes and their related hydrocarbons (isoalkanes, cycloalkanes, alkenes and carbonyl compounds) (Salzmann and Schiestl, 2007; Salzmann et al., 2007a, b; Scopece et al., 2007; Pellegrino et al., 2012; Naczk et al., 2018). This variability of floral odours has been thought to be adaptive and to reduce the avoidance learning of pollinators (Wright and Schiestl, 2009; Dormont et al., 2014). A few studies have compared the floral chemical signals of food-deceptive orchids with co-flowering rewarding species (emission by anthers, petals and/or nectar); the available data show possible similarities between them (Knudsen et al., 1993; Gumbert and Kunze, 2001; Johnson et al., 2003; Dötterl and Vereecken, 2010).

Dactylorhiza is a suitable model genus for investigation of the evolution of floral chemical signals. The monophyletic Dactylorhiza genus includes many species (approx. 75) (Pillon et al., 2006, 2007). Morphological, karyological and genetic studies may often fail to delineate taxa and their relationships, creating a situation in which almost no hypothesis explaining their origins can be rejected (Hedrén et al., 2001; Pillon et al., 2006). Auto- and allotetraploids have evolved several times by repeated gene duplication and, for the latter, by hybridization (Hedrén, 1996). We analysed and compared the chemical compounds in flowers of food-deceptive Dactylorhiza taxa: diploid D. incarnata sensu lato (s.l.) (L.) Soó including D. incarnata var. incarnata (Fig. 1A) and D. incarnata var. ochroleuca (Fig. 1B), as well as D. fuchsii (L.) Soó (Fig. 1C) and the allotetraploid descendant D. majalis complex (Fig. 1D). Hedrén et al. (2001, 2007) and Pedersen (2009) suggested that incarnata and ochroleuca flower morphs are best treated as varieties, implying that the concept of subspecies should be restricted to situations in which the taxa are more coherent, with constituent populations being more closely related to each other than to populations of other subspecies. In Europe, the D. majalis complex has repeatedly evolved by hybridization between two broadly defined lineages – D. incarnata and D. fuchsii – inheriting their plastid genomes from D. fuchsii as the maternal parent (Hedrén et al., 2007). The allotetraploid D. majalis complex lineage is genetically homogeneous but derived, including a variety of plastid haplotypes not encountered in the maternal parent. A phylogenetic history of the Dactylorhiza taxa based on chlorplast DNA (cpDNA) and internal transcribed spacer (ITS) markers separated the D. incarnata complex from D. fuchsii and D. majalis (Balao et al., 2016; Hedrén, 1996; Devos et al., 2006; Hedrén et al., 2001; Pillon et al., 2007). These Dactylorhiza taxa broadly share the same geographical distribution (Northern and Central Europe, and Western Asia) and often grow in sympatry (Tutin et al., 1980). All taxa have spurs, but possible nectar traces were rarely recorded (Naczk et al., 2018). Focusing on the above-mentioned properties of the studied Dactylorhiza, we applied genetic analyses using molecular cpDNA markers and flow cytometry to confirm the taxonomic positions of the investigated taxa, in particular the genetic similarity of D. incarnata var. incarnata and D. incarnata var. ochroleuca, and to exclude the occurrence of hybrids between them. Then we addressed three main questions: (1) whether these taxa produce shared and/or had taxon-specific combinations of floral chemical compounds; (2) which insects are attracted by the flowers of these taxa and are recognized as pollinators and/or flower visitors; and (3) what evolution model based on the interaction between pollinators and floral chemical compounds best explains deception in the genus Dactylorhiza.

Fig. 1.

Studied Dactylorhiza taxa and their pollinators: D. incarnata var. ochroleuca (A) and D. incarnata var. incarnata (B) with Apis mellifera on a flower (A. Wróblewska), D. fuchsii (C) with Strangalia attenuata (E. Brzosko) and D. majalis (D) with Apis mellifera (B. Ostrowiecka).

MATERIALS AND METHODS

Plant material

In 2016 we studied nine Dactylorhiza populations in different parts of north-eastern Poland (52°49’50’’N to 53°54’39’’N, 22°35’29.5’’E to 22°56’32’’E, Table 1, Fig. 2). We investigated one population consisting of D. incarnata var. incarnata individuals only (GO) and two populations with both D. incarnata var. incarnata and var. ochroleuca coexisting in sympatry in one community in the Biebrza Valley (ZB) and Rospuda Valley (RO; Table 1, Fig. 2). The populations in the Biebrza Valley and Rospuda Valley occupied sedge communities with low numbers of rewarding plant species such as Lychnis flos-cuculi and Ranunculus acris. Dactylorhiza fuchsii was observed in open hornbeam forest, also with a low share of rewarding plant species (<10 %) such as Anthriscus sylvestris, Stellaria nemorum and Geranium robertianum (Table 1). Three D. majalis populations (KA, SKI and SKII) and two D. fuchsii populations (CM and BR) were located in the Białowieża Primeval Forest and its vicinity, and one D. fuchsii population (GR) was in the Biebrza Valley. Dactylorhiza majalis grows in damp grassland with co-flowered Geum rivale, Ranunculus acris, Cardamine pratensis, Veronica chamaedrys and Alchemilla sp. The time of flowering of D. incarnata var. incarnata, D. incarnata var. ochroleuca and D. fuchsii differed from that of D. majalis. The first three taxa flowered between the beginning and end of June, whereas D. majalis flowers at the beginning of May and finishes flowering at the beginning of June.

Table 1.

Characteristics of nine Dactylorhiza populations in north-eastern Poland

Taxon	PC	GSP location	Habitat	Co-flowering rewarding plants	Pollinators*	Pollinators**	Flower visitors**
D. incarnata var. incarnata	GO	53°29’37.7’’N 22°45’29.7’’E	High frequency of grass and sedge	Lychnis flos-cuculi, Ranunculus acris, Geum rivale, Myosotis palustris, Galium aparine, Cerastium arvense	Hymenoptera: Apis mellifera1,11, Bombus pascuorum, B. pratorum, B. terrestis, B. lapidarius1,2, B. humilis10	Hymenoptera: Apis mellifera, Bombus sp.	Diptera: Chrysotoxum festivum Hymenoptera: Halictus rubicundus Coleoptera: Plateumaris sp., Oxythyrea funesta Lepidoptera: Lycaena sp.
D. incarnata var. incarnata and D. incarnata var. ochroleuca	ZB	53°17’59’’N 22°35’44’’E		Lysimachia thyrsiflora, Geum rivale, Galium ulginosum
D. incarnata var. incarnata and D. incarnata var. ochroleuca	RO	53°54’39’’N 22°56’32’’E		Lychnis flos-cuculi, Scutelaria galericulata, Stellaria graminea, Pedicularis palustris
D. fuchsii	CM	52°41’03’’N 23°39’07’’E	Open hornbeam forest with frequency <10 % of co-flowering rewarding plants	Anthriscus sylvestris, Stellaria nemorum, Crepis paludosa, Veronica chamaedrys, Ranunculus acris, Geranium robertianum	Hymenoptera: Apis mellifera, Bombus terrestis, B. pascuorum1,5,6,7 Diptera: Empis tessellata1,5,6 Coleoptera: Oedemera nobilis, O. lurida, Ruptela maculata1,5,6, Alossterna tabacicolor3, Cantharis fusca4, Paracorymbia maculicornis8, Pachytodes cerambyciformis3,8,9	Hymenoptera: Apis mellifera, Volucella pellucens, V. bombylans Coleoptera: Strangalia attenuata, Alosterna tabacicolor	Hymenoptera: Myrmica rufa Diptera: Phaonia sp., Culex sp. Helophilus sp. Coleoptera: Dasytes sp., Anaspis sp., Anoplodera rufipes
	BR	52°50’59’’N 23°53’40’’E		Stellaria nemorum, Veronica chamaedrys, Ranunculus lanuginosus, Geranium robertianum
	GR	53°36’28.1’’N 22°50’26.2’’E		Stellaria nemorum, Crepis paludosa, Ranunculus repens, Geranium robertianum, Maianthemum bifolium
D. majalis	KA	52°53’00’’N 23°40’29’’E	Damp, hay-growing madows	Ranunculus acris, Myosotis palustris, Lychnis flos-cuculi, Trifolium pratense, Cardamina pratensis	Hymenoptera: Apis mellifera1,14, Bombus pascuorum, B. pratorum, B. soroensis1, B. ruderarius3, Bombus sp.12,13,Eucera longicornis, Halictus leucozonius, Nomada sexfasciata14	Hymenoptera: Apis mellifera	Hymenoptera: Halictus sp., Lasioglossum sp., Lasioglossum majus Diptera: Helophilus pendulus, Chrysotoxum varnale, Pollenia sp., Scaeva pyrastri, Sphaerophoria scripta, Syrphus sp. Coleoptera: Mononychus punctumalbum, Dasytes sp. Lepidoptera: Vanessa cardui, Pieris sp., Adscita statices
	SK I	52°49’50’’N 23°43’10’’E		Geum rivale, Ranunculus acris, R. repens, Cardamine pratensis, Veronica chamaedrys, Alchemilla sp.
	SK II	52°50’4’’N 23°41’47’’E		Ranunculus acris, Geum rivale, Lychnis flos-cuculi, Cardamine pratensis, Trifolium pratense, Veronica chamaedrys, Lathyrus pratensis, Alchemilla sp.

PC, population code: D. incarnata var. incarnata population, GO; D. incarnata var. incarnata with D. incarnata var. ochroleuca, ZB and RO; D. fuchsii, CM, BR and GR; D. majalis, KA, SKI and SKII.

Pollinators and flower visitors: *based on literature data, **based on video observations in this study.

¹ Ruiz (2010); ²Vallius et al. (2013); ³Gutowski (1990); ⁴Claessens and Kleynen (2011); ⁵Vöth (1983); ⁶Démares (2000); ⁷Paulus (2005); ⁸Vöth (1999); ⁹Danesch and Danesch (1984); ¹⁰AHO (2005); ¹¹Bournerias and Prat (2004); ¹²Müller (1881); ¹³Hansen and Olesen (1999); ¹⁴Knuth (1899).

Fig. 2.

Sampling localities of nine Dactylorhiza populations in north-eastern Poland. GO – with D. incarnata var. incarnata only; ZB (DI/DO) and RO (DI/DO) – with D. incarnata var. incarnata and D. incarnata var. ochroleuca; CM (DF), BR (DF) and GR (DF) – with D. fuchsii; KA (DM), SKI (DM) and SKII (DM) – with D. majalis.

cpDNA and ITS analysis

The study used samples collected from all nine populations of Dactylorhiza taxa (ten samples from each population, including D. incarnata var. incarnata and D. incarnata var. ochroleuca; Table 1; Fig. 2). Leaves were sampled and dried in silica gel. Genomic DNA was extracted from dry leaf tissue with a Genomic Mini AX Plant kit (A & A Biotechnology). In total, 110 samples were analysed for three polymorphic cpDNA sequences found among the 15 cpDNA sequences investigated. These sequences included the intron of the chloroplast trnL (UAA) gene, amplified by primers c and d; the intergenic spacer between the trnL (UAA) 3′ exon and trnF (GAA), amplified by primers e and f of Taberlet et al. (1991); and the pbsC–trnK spacer of Demesure et al. (1996). The ITS region was amplified using primers 17SE (ACGAATTCATGGTCCGGTGAAGTGTTCG) and 26SE (TAGAATTCCCCGGTTCGCTCGCCGTTAC; Sun et al., 1994). PCR was performed as described by Gielly and Taberlet (1994). DNA sequencing was performed using BigDye Terminator V 3.1 (Applied Biosystems); following the manufacturer’s instructions, samples were run on an ABI 3130 Genetic Analyzer. Sequences were assembled and edited using BioEdit 7.04 (Hall, 1999). The sequences from cpDNA are deposited at the NCBI (accession no. MG924887–MG924890).

To investigate the relationships between haplotypes, phylogenetic analyses were carried out using a Bayesian approach implemented in BEAST 2.0 (Drummond et al., 2012), with Platanthera bifolia as outgroup (Inda et al., 2012). We performed this analysis using a Yule speciation prior, with the most likely model of sequence evolution obtained in jModelTest 0.1.1 (Posada, 2008). The model fit of the nucleotide substitution models was assessed with the Akaike information criterion (AIC). Simulations were run using Markov chain Monte Carlo (MCMC) for 10 000 000 generations, with a score of 10 000 in the BEAUTI program. Finally, we discarded the trees after a 10 % burn-in using TreeAnnotator 1.7.2. The tree was visualized with FigTree 1.3.1 (Rambaut, 2010).

Flow cytometry measurements

Genome size was estimated using fresh young leaves collected from the same individuals sampled for cpDNA and ITS analysis (including the collection of ten leaf fragments for D. incarnata var. incarnata and another ten for D. incarnata var. ochroleuca). Leaves of Pisum sativum ‘Set’ (2C = 9.11 pg; Sliwinska et al., 2005) were used as internal standard. Samples were prepared according to the procedure described by Jedrzejczyk and Sliwinska (2010). The plant tissue was chopped with a sharp razor blade in a plastic Petri dish with 1 mL of nuclei isolation buffer [2.5 mm MgCl₂·6H₂O, 85 mm NaCl, 0.1 m Tris, 0.1 % (v/v) Triton X-100, pH 7.0] supplemented with propidium iodide (PI; 50 μg mL^–1) and RNase A (50 μg mL^–1). The suspension of nuclei was passed through a 50 μm mesh nylon filter. For each sample, 7000–10 000 nuclei were measured using a Partec CyFlow SL Green flow cytometer (Partec GmbH, Münster, Germany) equipped with a high-grade solid-state laser, with green light emission at 532 nm and side (SSC) and forward (FSC) scatter. The analyses were performed for at least three replicates of plant tissue of each genotype. The obtained histograms were analysed using FloMax software (Partec, Münster, Germany). Nuclear DNA content values are presented as means and s.d. (in picograms, pg/2C). Genome size was estimated using the linear relationship between the ratio of target taxa and the internal standard 2C peak positions on the histograms. The mean coefficient of variation of 2C DNA content was estimated for all studied samples of Dactylorhiza taxa. 1C genome size was obtained by converting the values in picograms to base pair numbers using the factor 1 pg = 978 Mbp (Doležel et al., 2003).

To determine possible differences in nuclear DNA content among the studied genotypes, the results for nuclear DNA content were analysed using one-way analysis of variance (ANOVA) and Duncan’s test (P ≤ 0.05; Statistica v. 12, StatSoft Poland).

Identification of pollinators and flower visitors

For each Dactylorhiza population, two individuals with opened flowers (containing pollinaria) were videotaped (Sony HDR-CX410 AVCHD progressive, 8.9 MP) during five sunny, warm, windless days between 09.00 and 15.00 h. The cameras were set up on tripods 1–1.5 m from the plants to facilitate insect identification and to allow observation of pollinaria attachment to the insects. For each population, 60 h of data were recorded over the 2 year observation period. Here we focused only on taxonomical identification of the pollinators carrying away Dactylorhiza pollinaria, and also the flower visitors that were only landing on the flowers but not carrying pollinaria. We compared our data with data for Dactylorhiza pollinators from 14 published studies (Table 1).

Identification of floral chemical composition by gas chromatography–mass spectrometry

Five inflorescences from individuals previously assigned to the studied taxa using cpDNA, ITS and flow cytometry were collected from each of the nine populations (D. majalis in the middle of May; D. incarnata var. incarnata, D. incarnata var. ochroleuca and D. fuchsii in the middle of June). The flowers were then dried at room temperature for 24 h. Floral compounds were extracted with pentane and diethyl ether (3 mL per whole flower) at room temperature. The extraction was repeated three times, and all three extracts were combined. Then the material was transferred to a 5 mL vial, and 3 mL of pentane was added. Pentane (or hexane) is commonly used for extraction of scent from floral tissue (Ashman et al., 2005) and is an appropriate solvent for a broad range of plant volatiles (Prososki et al., 2007). The vial containing the sample and pentane was placed in an ultrasonic bath for 15 min. Next, the extract was filtered through a paper filter and transferred to a 10 mL vial, and the pentane was evaporated under a gentle stream of nitrogen at 30 °C. Finally the extract was redissolved in 100 μL of pentane and transferred to a chromatographic insert (200 μL capacity). The resulting mixture was subjected to gass chromatography–mass spectrometry (GC-MS) analysis. The residue (after extraction with pentane) was extracted three times with 3 mL of diethyl ether. The diethyl ether extract is less similar to the scent but more similar to non-volatile polar compounds. The extract was filtered through a paper filter and the solvent was removed with a gentle stream of nitrogen at 40 °C. The residue left on the wall was washed out with 3 mL of diethyl ether, and 0.5 mL of this solution was put into a 2 mL vial. After evaporation of the solvent, 450 μL of pyridine and 50 μL of N,O-bis(trimethylsilyl) trifluoroacetamide (BSTFA) were added to the vial. The reaction mixture was sealed and heated for 0.5 h at 60 °C to obtain trimethylsilyl (TMS) derivatives. The procedure was performed in triplicate (Ashman et al., 2005; Prososki et al., 2007).

The extracts were analysed on an Agilent 7890B gas chromatograph with an MSD 5977A mass selective detector fitted with an 7693A ALS autosampler system, electronic pressure control and split/splitless injector. Separation was performed on an HP-5ms (30 m × 0.25 mm, 0.25 μm film thickness) fused silica column at a helium flow rate of 1 mL min^–1. The injector worked in splitless mode at 250 °C. The EIMS spectra were obtained at 70 eV ionization energy, at a source temperature of 230 °C and quadrupole temperature of 150 °C. The MSD was set to scan at 33–620 amu. Chromatograms were registered in a linear temperature regime programmed from 50 to 320 °C at a rate of 3 °C min^–1. All the preparation steps and GC-MS analysis were performed for pure pentane and diethyl ether.

Identification in GC-MS analyses required two independent analytical parameters: mass spectrum and retention index (Van den Dool, 1963). Therefore a mixture of C₁₀–C₄₀ of n-alkanes (in hexane) was chromatographed to calculate the retention indices of chromatographic peaks. Experimental values of retention indices and mass spectra were compared with those from literature databases (NIST’14, Viley’9, Adams 4th), a home-made database (L.S.) and the Internet.

Chromatograms of all samples (pentane and ether extracts) were integrated to obtain areas for all chromatographic peaks, using ChemStation software (Agilent Technologies). The proportional distribution of each peak to total peak area was calculated. These values were used for further statistical analyses with R 3.4.0 (R Core Team, 2017). To visualize variation of the floral chemical components between samples of each taxon, dimension reduction with non-metric multidimensional scaling (NMDS) implemented in R vegan 2.4–3 was applied. To evaluate the effect of variation of emission rates among compounds on the grouping of samples, standard non-parametric analysis (Anderson, 2001) with permutational multivariate analysis of variance (PERMANOVA) was performed using Bray–Curtis dissimilarity and 4999 permutations, employing the vegan package in R (Oksanen et al., 2017; R Core Team, 2017). We tested whether there was an overall difference in floral chemical profiles between the studied taxa including two colour morphs, including the samples from all the populations; the Adonis function of the vegan package was used for this. To test the assumption of homogeneity of the dispersion functions, betadisper and permutes (4999 permutations) with ANOVA were used. To calculate pairwise contrasts between all sample pairs, the simper function (similarity percentage) from the vegan package was used (i.e. we determined which compounds were responsible for differences in the composition of volatiles between taxa).

RESULTS

Phylogenies of studied Dactylorhiza taxa

The final matrix consisted of concatenated sequences from trnL, trnL–trnF and pbsC–trnK and for ITS sequences for each Dactylorhiza taxon done separately. In each taxon, we identified only one cpDNA and ITS haplotype (the same haplotype for D. incarnata var. incarnata and D. incarnata var. ochroleuca); in total, three cpDNA and three ITS haplotypes were identified in the data set. The cpDNA and ITS haplotypes of D. fuchsii and D. majalis formed a well-supported lineage, and the second lineage was of D. incarnata var. incarnata and var. ochroleuca (100 % bootstrap support) (Supplementary data Fig. S1).

Nuclear DNA content and ploidy level

The genome size of the investigated taxa ranged from 6.74 pg/2C in D. fuchsii to 15.19 pg/2C in D. majalis, which corresponds to 6592 and 14 859 Mbp, respectively. The mean 2C DNA content of D. fuchsii was 6.90 pg. A similar genome size was obtained for D. incarnata var. incarnata (7.95 pg/2C) and D. incarnata var. ochroleuca (7.96 pg/2C). In D. majalis, the mean 2C nuclear DNA content was 15.16 pg. Intraspecific variation of genome size was observed only in D. fuchsii, in which the DNA content was highest in population GR (6.94 pg/2C) and lowest in population CM (6.74 pg/2C). The 2C DNA content differed by just 0.07 pg between populations RO and ZB of D. incarnata var. ochroleuca. Populations GO and ZB of D. incarnata var. incarnata showed the same nuclear DNA content (7.98 pg/2C), the largest genome size for this taxon. Among the three populations of D. majalis, genome size was highest in KA (15.19 pg/2C) and lowest in SK II (15.14 pg/2C) (Supplementary data Table S1; Fig. S1).

Pollinators and flower visitors of Dactylorhiza

We recorded seven insect taxa on D. incarnata. Apis mellifera and Bombus sp. (Hymenoptera) were categorized as pollinators (carrying out Dactylorhiza pollinaria). The remaining five taxa belonged to four groups (Hymenoptera, Diptera, Coleoptera and Lepidoptera) and were observed as flower visitors (not carrying out Dactylorhiza pollinaria; Table 1). On D. fuchsii we observed five pollinators: A. mellifera, Volucela pellucens and V. bombylans, belonging to Hymenoptera; and Strangalia attenuate (Fig. 1C) and Alosterna tabacicolor (Table 1), belonging to Coleoptera. Seven out of 12 insects (from Hymenoptera, Diptera and Coleoptera) were classified as flower visitors (Table 1). Dactylorhiza majalis flowers were visited by 16 insects, only one of which, A. mellifera (Hymenoptera; Fig. 1D), was a pollinator; the others were flower visitors (Hymenoptera, Diptera, Coleoptera and Lepidoptera) (Table 1).

To identify pollinators, we compared 14 published reports of insect visitation of D. incarnata, D. fuchsii and D. majalis (Table 1). These studies identified A. mellifera and multiple Bombus species (Hymenoptera) as shared pollinators of the Dactylorhiza taxa analysed here. Diptera and Coleoptera pollinators were also identified for D. fuchsii, and Hymenoptera pollinators for D. majalis (Table 1).

Floral chemical compounds

Twenty-two gas chromatograms were recorded for 11 samples (two replicates each): three each for D. majalis, D. fuchsii and D. incarnata var. incarnata, and two for D. incarnata var. ochroleuca. Two distinct sets of data came from the GC-MS measurements: chromatograms for the pentane extracts (more volatile, less polar compounds) and for the diethyl ether extracts (less volatile, more polar, TMS derivative compounds).

In the pentane extracts, we identified 81 volatile compounds (Supplementary data Table S2). Most of these were hydrocarbons (44 compounds, 62.2 % of the total peak area); among them, n-alkanes (38.4 %) had the largest peaks. The aldehydes were the second most abundant group of compounds, accounting for 17.9 % of the total peak. The other groups accounted for <3 % of the total area. Seventeen compounds occurred in all 11 samples. Thirty-seven compounds were common to all investigated taxa. We detected 77 compounds in D. incarnata var. incarnata, 71 in D. incarnata var. ochroleuca, 60 in D. fuchsii and 41 in D. majalis. Fifty-four compounds were identified in the ether extracts, the majority of which were benzenoids (24 compounds, 36.9 % of the total area) and non-aromatic acids (19 compounds, 13.4 % of the total area) (Supplementary data Table S3). Only three compounds were common to all samples: caproic acid, p-hydroxybenzyl alcohol and palmitic acid. The two most abundant compounds were p-hydroxybenzyl alcohol (20.6 %) and pelargonic acid (nonanoic acid, 5.4 %).

For pentane extracts, the metaMDS and ordiplot functions were used to visualize variation of emission rates (Fig. 3). The STRESS value was calculated at 0.08, indicating clear differences between the samples (STRESS <0.20). To confirm the above findings, PERMANOVA was performed using the adonis function [4999 permutations, n = 11, F(3,7) = 2.39, P = 0.005]. The results of our tests for homogeneity of variance by ANOVA [n = 11, F(3,7) = 0.29, P = 0.83] and permutest [4999 permutations, F(3,7) = 0.29, P = 0.84] indicated that the assumption of homogeneity was met.

Fig. 3.

Non-metric multidimensional scaling of the relative proportions of floral chemical compounds (extracted in pentane) in Dactylorhiza taxa: D. incarnata var. incarnata and D. incarnata var. ochroleuca, D. fuchsii and D. majalis, based on Bray–Curtis dissimilarity index [STRESS = 0.08; 4999 permutations; n = 11; F(3,7) = 2.39; P = 0.005].

For diethyl ether extracts, the groups of samples (investigated taxa including D. incarnata var. incarnata and D. incarnata var. ochroleuca) differed significantly [STRESS = 0.03 and PERMANOVA, 4999 permutations, n = 11, F(3,7) = 5.97, P = 0.019] but the tests for homogeneity of variance did not meet the assumptions [ANOVA, n = 11, F(3,7) = 5.88, P = 0.025; permutest, 4999 permutations, n = 11, F(3,7) = 5.88, P = 0.012].

DISCUSSION

The Dactylorhiza incarnata/maculata complex is an unusually dynamic system of polyploid speciation in which polyploids evolve continuously from the same set of broadly defined parental lineages (Hedrén, 2003). In Europe, the allotetraploid taxa, i.e. those belonging to D. majalis s.l., have evolved repeatedly by hybridization between the two parental lineages D. incarnata s.l. and D. maculata s.l. For a better understanding of the evolution of food deception in Dactylorhiza based on floral chemical compounds or other taxonomical traits, the taxonomic position and the genetic relationships between the diploid ancestors D. fuchsii and D. incarnata and their polyploid descendant D. majalis need to be confirmed. Our study employing a phylogenetic approach and flow cytometry confirmed previous reports (Pillon et al., 2006, 2007; Ståhlberg and Hedrén, 2008; Hedrén et al., 2012) that the investigated plants belong to three clearly distinguished Dactylorhiza lineages: D. fuchsii, D. incarnata and D. majalis, and we recorded no introgression or hybrids between them. However, we found support for the suggestion (Hedrén and Nordström, 2009) that D. incarnata with red and yellow flowers may best be recognized as var. incarnata and var. ochroleuca, respectively. This is the first survey showing the similarity of nuclear DNA content between D. incarnata var. incarnata and var. ochroleuca. The above-mentioned authors also suggested that subspecies level would be required in areas outside the range of other D. incarnata forms or in habitats where red/purple-flowered forms did not occur.

Besides the closer phylogenetic relationships between D. fuchsii and D. majalis than between these two taxa and D. incarnata inferred from plastid DNA, our study revealed that the floral chemical profiles of D. incarnata var. incarnata and D. incarnata var. ochroleuca are similar and that they differ from those of the other two taxa. The floral chemical components of Dactylorhiza have been studied in plants in a few populations of D. sambucina (Nilsson, 1980), D. romana (Salzmann and Schiestl, 2007), D. maculata (Schiestl and Cozzolino, 2008), D. majalis, D. incarnata and D. fuchsii (Naczk et al., 2018). We identified 112 floral chemicals in the studied Dactylorhiza taxa. The earlier Dactylorhiza surveys focused on identifying floral chemical compounds such as monoterpenes and n-alkanes/alkenes, as well as general variation of floral scent. Our findings are mostly consistent with those of previous studies and also indicate that multiple floral chemical strategies have evolved within Dactylorhiza. Our study also contributes evidence that different Dactylorhiza taxa produce similar floral chemical compounds, although their concentrations vary among them and their populations, and that taxon-specific compounds are also present. This suggests that they attract both shared and species-specific pollinators which play crucial roles in pollination of the studied Dactylorhiza, as was confirmed by our field video observations. We noted the pollinators Apis mellifera and Bombus in pure populations of D. incarnata var. incarnata and in mixed populations of D. incarnata var. incarnata and D. incarnata var. ochroleuca (Fig. 2A). The D. fuchsii pollinators were A. mellifera, Volucella pellucens, V. bombylans, Strangalia attenuata and Alosterna tabacicolor (Fig. 2B). The Coleoptera pollinators of D. fuchsii occur in the habitats of that species, particularly at the margins of moist deciduous mixed forests and woodland glades. The first individuals of these Coleoptera appear in June; we observed copulation of males and females of each Coleoptera species on D. fuchsii flowers and these individuals appeared to be the main active takers of D. fuchsii pollinaria. This phenomenon was also reported by Gutowski (1990) in the Białowieża Forest. The occurrence of these beetles is also associated with feeding (pollen and nectar) on Apiaceae and Rosaceae species, which occurs together with the studied populations of D. fuchsii; apparently they benefit from the same pollinators. On D. majalis, A. mellifera was the only pollinator (Fig. 2C). The three investigated Dactylorhiza orchids presumably can have similar taxonomic groups of pollinators within a given geographical region, but only A. mellifera was identified as a shared pollinator (B. Ostrowiecka et al., unpubl. research).

Based on our findings, we suggest that some of the floral chemical compounds in Dactylorhiza orchids can reduce species-specific pollination and promote hybridization between taxa. The identified floral compounds indicate that diverse pollinator-mediated chemical communication systems operate in Dactylorhiza. Unfortunately, few studies have investigated the pollinators and insect visitors of Dactylorhiza. Those studies identified both shared and species-specific pollinator taxa, and furnish partial support for our findings that the main pollinators of D. incarnata and D. majalis are Hymenoptera (A. mellifera, Bombus sp., Eucera longicornis, Halictus leucozonius, Nomada sexfasciata and Osmia fusca). Dactylorhiza fuchsii is also pollinated by Hymenoptera but mainly by A. mellifera and Bombus sp., as well as Coleoptera (Alosterna tabacicolor, Ruptela sp., Cantharis sp. and Oedemera sp.) and Diptera (Volucella) (Knuth, 1899; Gutowski, 1989; Ruiz, 2010; Claessens and Kleynen, 2011) (Table 1).

The floral chemical compounds of Dactylorhiza taxa and food-deceptive orchids have not yet been described comprehensively. Most studies have focused on their floral scent, which is composed mainly of fatty acid derivatives, benzoides, terpenoids, esters, alkenes and n-alkanes (Knudsen et al., 2001; Schiestl 2005; Salzmann and Schiestl, 2007; Salzmann et al., 2007a, b; Scopece et al., 2007; Schiestl and Cozzolino, 2008; Brodmann et al., 2009; Pellegrino et al., 2012; Dormont et al., 2014; Naczk et al., 2018). Some of these substances act as pheromones or affect the social and feeding behaviour of pollinators and visiting insects (Döttler and Vereecken, 2010). Generally, in Dactylorhiza we detected blends of chemical cues characteristic of both non-rewarding and rewarding plants (Knudsen et al., 2001). In the studied Dactylorhiza taxa, the bouquets of various n-alkanes at concentrations higher than the alkene concentrations may be key to the mechanism of chemical deception. The synergistic effect of n-alkanes and alkenes can increase the potency of the flower’s signal (Schiestl et al., 2000). In Orchidaceae, this phenomenon is usually described in sexually deceptive orchids (Schiestl et al., 1999; Schiestl and Cozzolino, 2008) but rarely in food-deceptive orchids. Cozzolino and Widmer (2005) suggested that these chemical components are also found in the food-deceptive plants from which sexually deceptive orchids probably evolved. Schiestl and Cozzolino (2008) reported a moderate amount of alkenes in D. maculata, explaining this as retention of the primitive state of alkene production. The n-alkanes and alkenes that are highly concentrated in Dactylorhiza flowers are worthy of attention, as they manipulate the behaviour of their pollinators in various ways. This manipulative role accounts for the typically abundant production of the alkanes tricosane, pentacosane, Z-(9)-tricosene and Z-(9)-pentacosene by these orchids; the latter two alkanes have been linked to waggle dancing in bees (Riley et al., 2005; Thom et al., 2007). During this dance, bees produce and release these alkanes to communicate information to other colony members about the direction and distance to patches of flowers yielding nectar and pollen, water sources or new nest site locations (Riley et al., 2005). 9-Heptacosene, the most common and most concentrated compound present in the studied Dactylorhiza taxa, plays a role in establishing the pheromonal dominance of reproductively active honey-bee workers in the colony (Okosun et al., 2015). Tergal gland alkenes from reproductive workers could act as releaser and primer pheromones in synergy with other glandular compounds to achieve pheromonal and reproductive dominance. Several of the detected hydrocarbons (e.g. hexadecane) are aromatic compounds found in environmental or food systems (Arn and Acree, 1998). The third important group of chemical signals detected in the investigated Dactylorhiza flowers was the aldehydes, dominated by nonanal, eicosanal and hexanal compounds. These interesting components are also recognized as sesquiterpenes (Knudsen et al., 2001), important components of floral odour in the appetitive proboscis extension response (PER) in honey-bees (Bitterman et al., 1983; Schiestl, 2010; Matsumoto et al., 2012; Wang et al., 2016).

Among the analysed Dactylorhiza taxa, D. majalis had the most distinct floral chemical profile. The flowers of D. majalis, which are the first to bloom among the studied Dactylorhiza taxa, contain more aliphatic carboxylic acid than the other investigated Dactylorhiza. This carboxylic acid is used as a precursor in the production of the esteric pheromone components of Bombus sp. and is known to be a pheromone of flies, bees (Ayasse et al., 2001; Devillers and Pham-Delégue, 2002) and beetles (Ratnayake et al., 2007). Surprisingly, foul-smelling compounds such as caproic, heptanoic, octanoic, palmictic and nonanoic acid were detected in flowers of D. majalis. These compounds are usually found in the faeces and urine of animals, especially after degradation by bacteria (Arnould et al., 1998; Smith et al., 2000), and in the floral scent of sapromyiophilous stapeliads such as two Echidnopsis species (Jürgens et al., 2006; Ollerton and Raguso, 2006). Flies are especially attracted to inflorescences that emit a pungent aroma; they can pollinate the flowers while visiting (Woodcock et al., 2014). Dressler (1981) and Pemberton (2010) found that some flies were strongly attracted to Orchidaceae flowers that emit odours of carrion and dung, deceptively attracting a particular guild of Diptera and Coleoptera that seek decaying organic matter for food and oviposition. Interestingly, during 2 years of observations, Diptera (e.g. Helophilus pendulus, Pollenia sp., Scaeva pyrastri, Sphaerophoria scripta and Syrphus sp.) were observed only as insect visitors and not as pollinators in the D. majalis populations in the vicinity of the Białowieża Primeval Forest (B. Ostrowiecka et al., unpubl. res.). Ratnayake et al. (2007) provided equivocal evidence suggesting that some floral chemical compounds act as beetle attractants by mimicking insect pheromones, such as the predominant saturated fatty acid, palmitic acid, which they found at relatively high concentrations in one population of D. incarnata var. ochroleuca and D. majalis.

The high amounts of alkenes and benzenoids produced by D. fuchsii and D. incarnata suggest that these compounds are used as attractants of Lepidoptera and Hymenoptera (Andersson et al., 2002; Dötterl et al., 2006). This type of p-hydroxybenzyl adduct (gastrodigenin) is common in orchids and is found mostly in the food-deceptive genera (Arundina, Bletilla and Pleione) (Majumder and Ghosal, 1993; Liu et al., 2004; Feng et al., 2008; Liu et al., 2009). A high concentration of p-hydroxybenzyl alcohol is common in flower tissue of those genera; in Dactylorhiza, such high levels were previously reported only in D. incarnata and D. majalis (Naczk et al., 2018). Kobayashi et al. (2003) reported that it was accumulated in carrot flowers and fruits (both ripe and unripe) but not in vegetative tissues. p-Hydroxybenzyl alcohol was also recognized as a volatile compound in Vanilla, describing it as vanilla like. This floral chemical was identified in orchids including Catasetum viridiflavum, Miltonia schroederiana (Kaiser, 1993), Epipactis (Jakubska, 2005) and Vanila planifolia (Podstolski et al., 2002). The benzoide p-coumaric acid is a major component of pollen grains and was abundant in the studied D. incarnata var. incarnata and D. incarnata var. ochroleuca. We note that p-coumaric acid is ubiquitous in the natural diet of honey-bees and might function as a nutraceutical, regulating immune and detoxification processes (Mao et al., 2013; Minh et al., 2016). Oldroyd (2007) suggested the absence of p-coumaric acid as a possible contributor to ‘colony collapse disorder’ in the honey-bee, because p-coumaric acid has been found to help honey-bees detoxify certain pesticides. Other benzoides detected in the above taxon Dactylorhiza, such as p-cresol (pungent aroma), are also typical constituents of dung (Kite et al., 1998), and their attractiveness to dung beetles has been demonstrated (Francke and Dettner, 2005). Caffeic acid is another interesting phenolic compound (similar to p-coumaric acid) observed in D. incarnata var. incarnata and D. incarnata var. ochroleuca. Although this secondary metabolite was described only in nectar, Hagler et al. (1993) and Wright et al. (2013) stressed that it manipulates pollinator behaviour; Apis spp. often forage low concentrations of phenolics, and honey-bees learn floral scents more effectively than those rewarded with sucrose alone. The D. fuchsii bouquet was also characterized by a high concentration of esters, particularly fatty acid esters (e.g. methyl hexadecanoate), which are also found in the pheromone gland of Coleoptera females (Reddy and Tangtrakulwanich, 2014).

Fig. 4.

Non-metric multidimensional plot of the relative proportions of floral chemical compounds (extracted in diethyl ether) in Dactylorhiza taxa: D. incarnata var. incarnata and D. incarnata var. ochroleuca, D. fuchsii and D. majalis, based on Manhattan dissimilarity index [STRESS = 0.03 and PERMANOVA, 4999 permutations; n = 11; F(3,7) = 5.97; P = 0.019].

Conclusions

Floral chemistry is hypothesized to be the product of natural selection, but researchers have just begun to consider the micro-evolution of these traits (Parachnowitsch, 2014). In food-deceptive orchids defined by Ackerman (1986), Cozzolino and Widmer (2005) and Jersáková et al. (2009), we can see the multidirectional character of the floral chemical signals. In this study we showed that the floral chemistry of deceptive Dactylorhiza includes both shared and taxon-specific chemical signals. We suggest that these floral chemical signals might have evolved through the action of two mechanisms. First, the basic floral chemical compounds (e.g. aldehydes and n-alkanes/alkenes), which are considered to be pheromones and which relay information about food (nectar and/or pollen), probably attract similar pollinators to each taxon. Secondly, the domination of single chemical groups in the different taxa (e.g. benzoids in D. incarnata and D. fuchsii, esters in D. fuchsii and acids in D. majalis) enriches the plant–insect chemical communication system, attracting different additional groups of pollinators. Interaction with pollinators in specific habitats would be expected to generate balanced selection for deceit pollination, leading to some intrataxon variation of floral chemistry. Further investigation of pollinator diversity in each studied Dactylorhiza population should be done in studies employing ecological approaches.

SUPPLEMENTARY DATA

Supplementary data are available online at https://academic.oup.com/aob and consist of the following. Figure S1: phylogenic relationship of chloroplast DNA (cpDNA, trnL, trnF, psbC–trnK sequences) and haplotypes and internal transcribed spacer haplotypes between D. incarnata var. incarnata and D. incarnata var. ochroleuca, D. fuchsii, D. majalis and outgroup species Platanthera bifolia estimated using the Bayesian approach in BEAST 2.0 and mean nuclear DNA content of taxa. Table S1: nuclear DNA content of Dactylorhiza incarnata as var. incarnata and var. ochroleuca morphs, D. fuchsii and D. majalis. Table S2: mean composition of floral chemical compounds (extracted from pentane) in nine populations of three taxa D. fuchsii, D. incarnata and D. majalis expressed as a percentage relative to total chemical compounds. Table S3: mean composition of floral chemical compounds (extracted from ether) in nine populations of three taxa D. fuchsii, D. incarnata and D. majalis expressed as a percentage relative to total chemical compounds.

FUNDING

This research was funded by a grant from the National Science Centre in Poland (no. 2013/09/B/NZ8/03350). The equipment of the BioNanoTecho Centre of the University of Bialystok (GC-MS) was supported in part by European Regional Development Fund (project no. POPW.01.03.00-20-004/11, 2007–2013).