The Chemical Basis of Host-Plant Recognition in a Specialized Bee Pollinator
Paulo Milet-Pinheiro, Manfred Ayasse, Heidi E. M. Dobson, Clemens Schlindwein, Wittko Francke & Stefan Dötterl Journal of Chemical Ecology ISSN 0098-0331 Volume 39 Combined 11-12 J Chem Ecol (2013) 39:1347-1360 DOI 10.1007/s10886-013-0363-3
1 23
Your article is protected by copyright and all rights are held exclusively by Springer Science +Business Media New York. This e-offprint is for personal use only and shall not be selfarchived in electronic repositories. If you wish to self-archive your article, please use the accepted manuscript version for posting on your own website. You may further deposit the accepted manuscript version in any repository, provided it is only made publicly available 12 months after official publication or later and provided acknowledgement is given to the original source of publication and a link is inserted to the published article on Springer's website. The link must be accompanied by the following text: "The final publication is available at link.springer.com”.
1 23
Author's personal copy J Chem Ecol (2013) 39:1347–1360 DOI 10.1007/s10886-013-0363-3
The Chemical Basis of Host-Plant Recognition in a Specialized Bee Pollinator Paulo Milet-Pinheiro & Manfred Ayasse & Heidi E. M. Dobson & Clemens Schlindwein & Wittko Francke & Stefan Dötterl
Received: 5 April 2013 / Revised: 26 August 2013 / Accepted: 15 October 2013 / Published online: 14 November 2013 # Springer Science+Business Media New York 2013
Abstract Many pollinators specialize on a few plants as food sources and rely on flower scents to recognize their hosts. However, the specific compounds mediating this recognition are mostly unknown. We investigated the chemical basis of host location/recognition in the Campanula -specialist bee Chelostoma rapunculi using chemical, electrophysiological, and behavioral approaches. Our findings show that Ca. trachelium flowers emit a weak scent consisting of both widespread and rare (i.e., spiroacetals) volatiles. In electroantennographic analyses, the antennae of bees responded to aliphatics, terpenes, aromatics, and spiroacetals; however, the bioassays revealed a more complex response picture. Spiroacetals attracted host-naive bees, whereas spiroacetals together with aliphatics and terpenes were used for host finding by host-experienced bees. On the intrafloral level, different flower parts of Ca. trachelium
showed differences in the absolute and relative amounts of scent, including spiroacetals. Scent from pollen-presenting flower parts elicited more feeding responses in host-naive bees as compared to a scentless control, whereas hostexperienced bees responded more to the nectar-presenting parts. Our study demonstrates the occurrence of learning (i.e., change in the bee’s innate chemical search-image) after bees gain foraging experience on host flowers. We conclude that highly specific floral volatiles play a key role in hostflower recognition by this pollen-specialist bee, and discuss our findings into the broader context of host-recognition in oligolectic bees.
Keywords Chelostoma . Campanula . Floral scents . Host location/recognition . Oligolectic bees . Spiroacetals
In memoriam Jan Tengö Electronic supplementary material The online version of this article (doi:10.1007/s10886-013-0363-3) contains supplementary material, which is available to authorized users. P. Milet-Pinheiro : M. Ayasse Institute for Experimental Biology, University of Ulm, 89069 Ulm, Germany P. Milet-Pinheiro (*) Laboratório de Ecologia Química, Departamento de Química Fundamental, Universidade Federal de Pernambuco, 50740-560 Recife, Brazil e-mail:
[email protected]
W. Francke Institut für Organische Chemie, University of Hamburg, 20146 Hamburg, Germany
S. Dötterl Department of Plant Systematic, University of Bayreuth, 95440 Bayreuth, Germany
H. E. M. Dobson Department of Biology, Whitman College, Walla Walla, WA 99362, USA C. Schlindwein Departamento de Botânica, Universidade Federal de Minas Gerais, 31270-901 Belo Horizonte, Brazil
Present Address: S. Dötterl Organismic Biology, Plant Ecology, University of Salzburg, 5020 Salzburg, Austria
Author's personal copy 1348
Introduction Floral signals generally are essential in any plant-pollinator interaction, and often consist of combinations of visual cues (e.g., color and shape) and olfactory stimuli (scents, derived for example from petals or pollen) that elicit responses in the pollinators (Chittka and Raine 2006; Dobson and Bergström 2000; Lunau 1995; Raguso 2008). Plants associated with specialized pollinators must be characterized by specific floral signals that allow unambiguous identification of the host plants by these pollinators. Although visual cues are important in attracting pollinators, they often are not as specific as olfactory cues (Kunze and Gumbert 2001; Wright and Schiestl 2009). Floral scents have the advantage in that they can provide a species-specific identity to flowers with their potentially infinite diversity (Knudsen et al. 2006; Williams 1983). In fact, highly specific compounds or unique ratios of common compounds have been shown to allow host-flower recognition in specialized and uncommon interactions, such as nursery (larvae of pollinators feed on floral tissue; Chen et al. 2009; Dötterl et al. 2006; Grison-Pigé et al. 2002; Svensson et al. 2010) and sexually deceptive orchid (Ayasse et al. 2003; Peakall et al. 2010; Schiestl et al. 1999; Schiestl and Ayasse 2001; Schiestl and Peakall 2005) pollination systems. Specialization among bees may reflect the spectrum of plant species they use as pollen source. In contrast to the highly generalized honeybees, many solitary wild bee species (15–60 % of the total bee fauna depending on the locality, Minckley and Roulston 2006) specialize on certain plants within a genus or family for collecting pollen, which they use to rear their offspring (Cane and Sipes 2006; Robertson 1925; Müller and Kuhlmann 2008). Currently, our empirical understanding of the chemical bases governing host-plant location in bees, including both pollen-generalist (polylectic) and pollen-specialist (oligolectic) species, is still in its infancy (Dötterl and Vereecken 2010). For oligolectic bees, only four studies have been conducted to investigate the role of floral scent components (either individually or as a blend) in hostplant location or recognition (Andrews et al. 2007; Burger et al. 2012; Dobson and Bergström 2000; Dötterl et al. 2005a). In these studies, the compounds eliciting behavioral responses in the bees were shown to be either unusual, such as protoanemonine (Dobson and Bergström 2000) and benzoquinone (Burger et al. 2012), or widespread ( Andrews et al. 2007; Dötterl et al. 2005a). However, only Burger et al. (2012) identified chemicals exclusive to the host flowers that could be used for discriminating host from non-host plants. Therefore, the chemical basis of host-plant location is little understood in most oligolectic bees and awaits further experimental investigations. An interesting tendency emerging from the studies cited above is that the chemical basis of host location in oligolectic
J Chem Ecol (2013) 39:1347–1360
bees changes with a bee’s foraging experience: bees rely initially (i.e., as host-naive bees, immediately after emergence) on compounds other than those used after they have visited the flowers (i.e., become host-experienced bees). Such experience-based changes in the bee’s preference for certain olfactory floral cues appear to increase the efficiency of host location/recognition, which in turn suggest some parallels to the associative learning of floral cues in flower-visiting insects vis-a-vis rewarding and non-rewarding flowers, which likewise leads to increased foraging efficiency for pollen and nectar (Dobson 2006; Lunau and Maier 1995; Weiss 1991). Chelostoma rapunculi (Lepeletier 1841) (Fig. 1) is a European oligolege specialized on the pollen of bellflowers (Campanula species), and in the presence of host plants, both males and females restrict nectar gathering to these plants (Milet-Pinheiro et al. 2012; Westrich 1989). This bee frequently has been documented to be one of the effective pollinators of several Campanula species, together with other generalized social species, such as bumble bees and honey bees, and various solitary bees, including oligoleges of the genus Chelostoma, Andrena, and Melitta (Blionis and Vokou 2001; Schlindwein et al. 2005). Recently, it was shown that host-naive bees of Ch. rapunculi are innately attracted to the visual cues of one of its hosts, Campanula trachelium, as well as to the olfactory ones, and that the bees clearly prefer inflorescence cues of this host plant over those of non-host plants (Milet-Pinheiro et al. 2012). However, the origin (flower part) and chemical identity of the floral volatiles involved in the attraction of Ch. rapunculi are unknown. To address this gap in our understanding of the chemical ecology of host-plant recognition by this specialized pollinator, we performed a comprehensive multifaceted study combining floral scent chemical analyses, electrophysiological tests, and behavioral assays utilizing both natural floral scents and synthetic substances. We tested the hypothesis that Ch. rapunculi uses either highly specific and uncommon floral volatiles or specific ratios of more widespread compounds to recognize host flowers of Ca. trachelium , and addressed the following questions: 1) What floral compounds of Ca. trachelium attract Ch. rapunculi bees? 2) Are there differences in scent among floral parts, and which parts elicit landing and feeding responses in the bees? 3) Is the olfactorybased attraction the same in host-naive and host-experienced bees, or does the bee’s olfactory search-image of Ca. trachelium change with foraging experience and thus learning?
Methods and Material Sampling of Floral Volatiles Using standard dynamic headspace methods (Dötterl et al. 2005b), volatiles were collected from flowers of Ca. trachelium for three separate purposes: 1)
Author's personal copy J Chem Ecol (2013) 39:1347–1360
1349
To control for non-floral (vegetative) volatiles and contaminants in the floral scent samples, headspace samples of nonflowering plants (N =3) and blank controls (empty oven bags; N =3) also were collected following the same methods described above. All headspace samples were stored in screw cap vials at −20 °C until chemical and electrophysiological analyses.
Fig. 1 Chelostoma rapunculi female collecting pollen from Campanula trachelium. Note that bees brush the hairy style on which pollen is secondarily presented
to obtain scent samples for electrophysiological investigations; 2) to identify the components of the scent bouquet;, and 3) to determine how the scent varies among the different parts within a flower. Plants were cultivated to seedling stage in the greenhouse of the Department of Plant Systematics, University of Bayreuth, Germany, and subsequently outdoors in flowerbeds, where they grew to maturity. To obtain floral scent samples for the chemical (N =6) and electrophysiological (N =6) analyses, fresh inflorescences from three different plants (about 80 flowers total) were enclosed in a polyester oven bag (20×30 cm; Toppits®). Volatiles were trapped for 4 h in an adsorbent tube, through which air was drawn at a rate of 200 ml/min using a membrane pump (G12/01 EB, Rietschle Thomas, Puchheim, Germany). The adsorbent tubes consisted of ChromatoProbe quartz microvials (GC/MS: length: 15 mm; inner diam: 2 mm; Varian Inc., Palo Alto, CA, USA) or ChromatoProbe Pyrex glass vials (GC/EAD: length: 15 mm; inner diam: 1.5 mm; Aviv Analytical LTD, Hod Hasharon, Israel), cut at the closed end, and filled with a mixture of 1.5 mg Tenax-TA (mesh 60–80; Supelco, Bellefonte, PA, USA) and 1.5 mg Carbotrap B (mesh 20–40, Supelco, Bellefonte, PA, USA), which was held in the tubes with glass wool (Dötterl and Jürgens 2005). To investigate the spatial scent pattern within flowers, fresh flowers at the male stage (first day of anthesis) were cut into separate parts. In open flowers, pollen is located mainly on the style, and the dehisced stamens shrivel, since Campanula flowers have secondary pollen presentation (Jost 1918; Kirchner 1897), where the pollen is deposited along the length of the style prior to anthesis. The experimental flowers were divided into three parts for scent sampling: 1) corolla (fused petals); 2) pollen + style + stamens (excluding the filament bases, which form flaps covering the nectaries); and 3) inferior ovary + nectaries (including the filament bases). Scent samples (N =6) were collected by enclosing flower parts of 15–34 flowers (exact number depended on the availability of malestage flowers) in oven bags (10×15 cm) for 1 h (to saturate the air), and then collecting volatiles for 2 min into adsorbent tubes (see above).
Chemical Analysis To identify volatiles in the floral scents of Ca. trachelium, headspace samples were analyzed on a Varian Saturn 2000 mass spectrometer coupled to a Varian 3800 gas chromatograph (GC/MS) equipped with a 1079 injector (Varian Inc., Palo Alto, CA, USA), which had been fitted with the ChromatoProbe kit (see Dötterl and Jürgens 2005). A quartz microvial was loaded into the probe, which then was inserted into the modified GC injector. The injector split vent was opened, and the injector was heated to 40 °C to flush any air from the system. After 2 min, the split vent was closed, and the injector heated to 200 °C/min, then held at 200 °C for 4.2 min, after which the split vent was opened and the injector cooled down. Separations were achieved with a fused silica column ZB-5 (5 % phenyl polysiloxane; 60 m long, inner diam 0.25 mm, film thickness 0.25 μm, Phenomenex). Electronic flow control was used to maintain a constant helium carrier gas flow of 1.8 ml/min. The GC oven temperature was held for 7 min at 40 °C, then increased by 6 °C per min to 250 °C and held for 1 min. The MS interface worked at 260 °C and the ion trap at 175 °C. Mass spectra were taken at 70 eV (in EI mode) with a scanning frequency of 1 scan/sec from m/z 30 to 350. The GC/MS data were processed using the Saturn Software package 5.2.1. Tentative identification of most of the compounds was carried out using the NIST 08, Wiley 7, and Adams (Adams 2007) mass spectral data bases, or the data base provided in MassFinder 3, and confirmed by comparison of retention times with published data. Structure assignment of spiroacetals was based on their mass spectrometric fragmentation pattern and on published mass spectra (Francke and Kitching 2001;Tengö et al. 1982). Structure assignments of individual components were confirmed by comparison of mass spectra and GC retention times with those of commercially available standards or compounds synthesized in our lab (for details on structure elucidation and synthesis of spiroacetals, see Electronic Supplementary Material – ESM 1). To quantify the absolute amount of each floral volatile in a sample, known amounts of monoterpenes, fatty acid derivatives, and aromatics were injected into the GC/MS system, and their mean peak areas were used to determine the total amount of each compound (for more details see Dötterl et al. 2005b). To determine which components of the samples were exclusively emitted by flowers, the composition of volatiles collected from blooming plants was compared with that from
Author's personal copy 1350
vegetative parts. Volatiles detected in the empty bags were considered ambient contaminants and were omitted from the experimental samples. Electrophysiology Electroantennografic detection coupled with gas chromatography (GC/EAD) was conducted on Ch. rapunculi to determine the bee’s antennal perception of compounds in the floral scent of Ca. trachelium. This analysis was performed on a gas chromatograph (HP 5890 Series 2, Hewlett-Packard, Palo Alto, CA, USA) equipped with a FID and a ZB-5 column (30 m long, 0.32 mm i.d., 0.25 μm film thickness, Phenomenex) coupled to an EAD setup (heated transfer line, two-channel universal serial bus acquisition controller) provided by Syntech (Kirchzarten, Germany). ChromatoProbe microvials containing trapped scent samples were inserted splitless by using the ChromatoProbe kit (AvivAnalytical, Hod Hasharon, Israel), which was attached to the injector, at an oven temperature of 60 °C and an injector temperature of 220 °C, followed by opening the split valve after 1 min and increasing the oven temperature at a rate of 10 C/min to 220 °C, which was held for 5 min. The effluent was split [using a four-arm splitter (GRAPHPACK 3D/2, Gerstel, Mülheim, Germany)], and 16 ml/min of make-up gas (helium) were added. The outlet of the EAD was placed in a cleaned and humidified airflow that was directed over the antenna of the bee. For the EAD, antennae were cut at their base and tip, and mounted between two electrodes, which were filled with insect ringer (8.0 g/l NaCl, 0.4 g/l KCl, 0.4 g/l CaCl2) and connected to silver wires. Electrophysiological measurements were performed separately with the antennae of six adult female bees, 4–7 d-old. A floral scent compound was considered to be EAD-active when it elicited a depolarization response in at least three of the six replicates.
Behavioral Experiments Attractiveness of EAD-Active Compounds and Inflorescences To establish whether the EAD-active compounds also trigger behavioral responses in Ch. rapunculi, mixtures with these compounds were prepared and tested on bees (males and females). From early June to end July, we performed behavioral experiments with host-naive and host-experienced bees in a large flight cage (7.2×3.6×2.2 m) in Bayreuth, the same described by Dötterl and Schäffler (2007). All bees used in the tests emerged within the flight cage from trap nests (cut stems of Phragmites australis, Poaceae), and were provided both sponge feeders saturated with sugar water (30 %, fructose and glucose 1:1) and non-host nectar flowers (Geranium pratense, Geraniaceae; Lythrum salicaria , Lythraceae; and Sinapis arvensis, Brassicaceae). Thus, the host-naive bees had foraging experience with non-host flowers, but not Campanula
J Chem Ecol (2013) 39:1347–1360
host-flowers. The bioassays were conducted first on hostnaive bees; when they were completed, flowering Ca. trachelium plants were placed in the cage to allow visitation by bees, and after 3 d these same bees were considered hostexperienced. Attraction of host-naive and host-experienced bees (males and females) to the EAD-active compounds was tested in dual-choice bioassays: odors were placed in two black quartz glass cylinders (29 cm height, 10 cm diam) with 60 small holes (diam 0.2 cm) (see also Burger et al. 2010b). Scented air inside the cylinders was blown out through the holes by a membrane pump (G12/01 EB, Rietschle Thomas, Puchheim, Germany) with a flow rate of 1 l/min. The paired cylinders in each bioassay were placed 1.5 m apart within the large flight cage. Each bioassay lasted 30 min; after the first 15 min, the position of the two cylinders was exchanged. Bioassays were conducted on sunny days between 10:00 and 15:00 h, when the bees were most active. Before each bioassay, all plants were removed from the flight cage early in the morning; after the test, they were placed back into the cage. Only flights toward cylinders and to a distance of 5 cm or closer were recorded as responses. During each bioassay, all responding bees were caught with insect nets and stored in an ice box to avoid pseudo-replication (counting an individual bee more than once). All caught bees were released back into the flight cage at the end of the bioassay and could participate in subsequent ones. Bioassays that did not provide a sufficient number of bee responses for statistical tests were repeated a second time and, to avoid recounting the same bees, all bees caught during the first bioassay were marked with permanent color markers before being released back into the cage. Due to stochastic factors such as mortality, the number of bees present in the flight cage varied among experiments (60 to 100 total individuals, males and females). Eight dual-choice bioassays were performed with mixtures of synthetic compounds identified in Ca. trachelium inflorescence scents as EAD-active on Ch. rapunculi: 1) mix of all EAD-active compounds (hereafter referred to as the “complete mix”) vs. control; 2) spiroacetals vs. control; 3) terpenes vs. control; 4) aromatics vs. control; 5) aliphatics (hereafter referred to as 2-nonanone) vs. control; 6) complete mix vs. spiroacetals; 7) complete mix vs. flowering stems of Ca. trachelium; and 8) spiroacetals vs. flowering stems. For the bioassays, a filter paper was impregnated with 10 μl of the test mixture diluted in acetone (release rate equivalent to the amount of scent emitted by 80 flowers; for more details see ESM 2). For the negative control, a filter paper was impregnated with 10 μl of acetone. For the positive control, we used flowering stems with a total of only eight flowers (the approximate number that could fit inside the cylinder), which is 10fold less than the number of flowers represented in 10 μl of the synthetic mixtures; therefore, when testing synthetic mixtures against the positive control, we used only 1 μl of the test
Author's personal copy J Chem Ecol (2013) 39:1347–1360
mixture, which is equivalent to eight flowers (see ESM 2). Although changes in scent pattern after tissue damage might be expected (see herbivore-induced volatiles; Kessler and Heil 2011), we have strong indications that this did not affect the attractiveness of Ca. trachelium flowers to Ch. rapunculi bees. In our flight cage, we frequently provided bees with flowering stems in water pots, and they promptly collected pollen and nectar on them, even if potted plants were present. Attractiveness of Scent from Different Flower Parts Differential attractiveness of the scents from separate parts of Ca. trachelium flowers was tested in multiple-choice behavioral experiments, using both host-naive and hostexperienced female bees, from mid-June to mid-July, at the research facility of Station Linné in Öland, Sweden. Hostnaive bees were obtained from wood trap-nests retrieved from the field; these were put into plexiglass cages (51×21×31 cm) with screen lids for bee emergence. Host-experienced bees were collected in the field while foraging on Ca. trachelium. All bees were marked with color dots using Liquid Paper®. Host-naive and host-experienced bees were kept in separate plexiglass cages and were fed exclusively sugar water (1:1 ratio, v/v), offered on Whatman filter paper placed in glass vials filled with sugar solution. Thus, no fresh flowers were provided to either group of bees. The same bioassay design was used for both host-naive and host-experienced bees. During each bioassay, bees were offered a choice of four samples: three scented samples corresponding to the separate flower parts used in scent analyses, namely 1) corolla, 2) pollen + style + stamens, and 3) inferior ovary + nectaries, and a fourth empty negative control. The samples were arranged equidistantly (16 cm apart) around the inner periphery of a small round mesh cage (24 cm high, 24 cm diam), in which bees could fly freely. Samples consisted of small glass vials (6 cm high, 3 cm diam) covered with dyed cotton cheesecloth: two layers of yellow cloth covered the vial opening and prevented the bee from contacting the flower material inside, and a paper skirt covered with blue cloth (same reflective wavelength as Ca. trachelium flowers) surrounded the vial top, creating a shallow, upward-facing, bell-shaped artificial flower resembling Campanula flowers. Scent samples were prepared from flowers of Ca. trachelium freshly gathered in the field. Each scent vial contained flower parts from 10 flowers (5 at the male stage, when pollen is provided, and 5 at female stage, when most nectar is produced, H. Dobson pers. obs.). Flowers were dissected by sequentially cutting off the 1) corolla and 2) style and five withered stamens (leaving in place the filament bases covering the nectaries); remaining was 3) the inferior ovary + nectaries. Newly prepared vials were allowed to become scent-saturated for 15 min prior to testing. Flower samples were replaced with fresh ones every four bioassays (approximately every 90 min).
1351
Bioassays were conducted outdoors on sunny days, between 8:00 and 15:00 h, at ambient temperature (≥ 18 °C). Each female bee (total bees tested: 55 host naive and 72 host experienced) was tested individually for 16 min; the cage was rotated every 2 min to even out impact from sunlight and wind. Bee responses to the samples were classified as: 1) landing: a flying bee lands on a sample; 2) feed-landing: a landing bee attempts to feed by extending its proboscis; and 3) active feed-landing: a feed-landing bee inserts its head or more of its body into the cheesecloth covering; these categories were non-exclusive. Any bee that did not attempt to feed was retested (up to a maximum of three times) until it displayed a feed-landing. Statistical Tests To test for multivariate differences in scent among flower parts, ANOSIM was used in Primer 6.1.6 (Clarke and Gorley 2006); the Bray-Curtis quantitative similarity index was calculated using Primer to assess pair wise semi-quantitative similarities between different parts based on the relative amounts of compounds (the absolute total amount was set as 100 %). Differences in absolute amount of emitted scent among the three flower parts was tested using FriedmanANOVA; Wilcoxon matched pairs test was used for post-hoc comparisons (StatSoft 2004). In the dual-choice bioassays, one-tailed exact binomial tests were used to test the null hypothesis that synthetic mixtures attract ≤ number of bees than acetone controls (one-tailed because it is unlikely that compounds in Ca. trachelium floral scent have repellent properties to Ch. rapunculi). Two-tailed exact binomial tests were used to test the hypothesis that the complete mix attracts the same number of bees as either the spiroacetal mix or flowering stems, and that the spiroacetals attract the same number of bees as flowering stems. Binomial tests were calculated using the spreadsheet provided by http://udel.edu/*mcdonald/ statexactbin.html (accessed 10 August 2012; see also McDonald 2009). Responses of male and female bees were pooled, as individuals of both sexes responded equally in all bioassays (Fisher’s exact tests: 0.13
0.35), but both emitted several fold higher
amounts than the corolla (P
