Evolution, 54(6), 2000, pp. 1995±2006

EVOLUTION OF REPRODUCTIVE STRATEGIES IN THE SEXUALLY DECEPTIVE ORCHID SPHEGODES: HOW DOES FLOWER-SPECIFIC VARIATION OF ODOR SIGNALS INFLUENCE REPRODUCTIVE SUCCESS?

MANFRED AYASSE,1,2 FLORIAN P. SCHIESTL,1 HANNES F. PAULUS,1 CHRISTER LOÈ FSTEDT,3 BILL HANSSON,3 FERNANDO IBARRA,4 AND WITTKO FRANCKE4 1Institute of Zoology, University of Vienna, Althanstrasse 14, A-1090 Vienna, Austria 2E-mail: [email protected] 3Department of Ecology, Lund University, S-22362 Lund, Sweden 4Institute of Organic Chemistry, D-20146 Hamburg, Germany

Abstract. The orchid Ophrys sphegodes Miller is pollinated by sexually excited males of the solitary bee nigroaenea, which are lured to the ¯owers by visual cues and volatile semiochemicals. In O. sphegodes, visits by pollinators are rare. Because of this low frequency of pollination, one would expect the evolution of strategies that increase the chance that males will visit more than one ¯ower on the same ; this would increase the number of pollination events on a plant and therefore the number of seeds produced. Using gas chromatography±mass spectrometry (GC-MS) analyses, we identi®ed more than 100 compounds in the odor bouquets of labellum extracts from O. sphegodes; 24 compounds were found to be biologically active in male olfactory receptors based on gas chromatography with electroantennographic detection (GC-EAD). Gas chromatography (GC) analyses of odors from individual ¯owers showed less intraspeci®c variation in the odor bouquets of the biologically active compounds as compared to nonactive compounds. This can be explained by a higher selective pressure on the pollinator-attracting communication signal. Furthermore, we found a characteristic variation in the GC-EAD active esters and aldehydes among ¯owers of different stem positions within an in¯orescence and in the n-alkanes and n-alkenes among from different populations. In our behavioral ®eld tests, we showed that male bees learn the odor bouquets of individual ¯owers during mating attempts and recognize them in later encounters. Bees thereby avoid trying to mate with ¯owers they have visited previously, but do not avoid other ¯owers either of a different or the same plant. By varying the relative proportions of saturated esters and aldehydes between ¯owers of different stem positions, we demonstrated that a plant may take advantage of the learning abilities of the pollinators and in¯uence ¯ower visitation behavior. Sixty-seven percent of the males that visited one ¯ower in an in¯orescence returned to visit a second ¯ower of the same in¯orescence. However, geitonogamy is prevented and the likelihood of cross-fertilization is enhanced by the time required for the pollinium deposited on the pollinator to complete its bending movement, which is necessary for pollination to occur. Cross-fertilization is furthermore enhanced by the high degree of odor variation between plants. This variation min- imizes learned avoidance of the ¯owers and increases the likelihood that a given pollinator would visit several to many different plants within a population.

Key words. Andrena nigroaenea, chemical communication, learning behavior, Ophrys sphegodes, pollination by sexual deception, reproductive success, scent variation.

Received September 14, 1999. Accepted June 4, 2000.

Pollination by sexual deceit is exclusive to the is pollinated by males of usually only one or a few pollinator (Ackerman 1986; Nilsson 1992), and has been documented species (Kullenberg 1961, 1973; Paulus and Gack 1986, in orchids growing in Australia (Peakall 1990), South Amer- 1990, 1994). Because Ophrys orchids are usually interfertile, ica (van der Pijl and Dodson 1966) and Europe (Kullenberg the species-speci®c attraction of pollinators is important for and BergstroÈm 1976; Paulus and Gack 1990). Orchids of the the reproductive isolation of each species. genus Ophrys grow in Europe, especially around the Medi- Ophrys ¯owers tend to produce complex bouquets of vol- terranean, and are pollinated by male insects, mostly bees atiles typically consisting of more than 100 species-speci®c (Andrenidae, Anthophoridae, Colletidae, Megachilidae, and chemical compounds (Borg-Karlson et al. 1985, 1987; Borg- Apidae), but occasionally by predatory (Sphecidae) and par- Karlson 1987, 1990). Behavioral experiments with synthetic asitic wasps (Scoliidae) and in a few cases by beetles (Scar- copies of the compounds produced by Ophrys ¯owers have abaeidae; Kullenberg 1961, 1973; Borg-Karlson 1990; Paulus shown that only certain volatiles are active in stimulating and Gack 1990). Male insects are lured to the orchid by visual mating behavior in the males (Kullenberg and BergstroÈm cues and volatile semiochemicals. At close range, chemical 1976; TengoÈ 1979; Borg-Karlson 1990). Among the com- signals from the ¯owers elicit sexual behavior in males, pounds produced by O. sphegodes, 14 hydrocarbons that were which respond in a manner similar to that shown to sex pher- identi®ed in similar proportions in cuticle extracts of the omones of females, whereby the males try to copulate with pollinators' females elicited copulation attempts in the males the ¯ower labellum. During these so-called pseudocopula- (Schiestl et al. 1999). Therefore, not all of the compounds tions, a male touches the gymnostemium of the ¯ower and may have a function in attracting male pollinators. The role the pollinarium may become attached to the insect's head or of those compounds that do not trigger copulatory behavior in some species to the tip of the abdomen; pollination takes in the males remains unknown. place when the male visits another ¯ower. Because ¯ower visits in non-food-rewarding pollination Numerous behavioral tests have shown that the Ophrys- systems are rare and brief (Peakall and Beattie 1996), food pollinator relationship is highly speci®c: Each Ophrys species and sexual deceptive plant species often show low levels of 1995 ᭧ 2000 The Society for the Study of Evolution. All rights reserved. 1996 MANFRED AYASSE ET AL. fruit set compared to food-rewarding species (Hutchings al. 1989) and P. bifolia (Tollsten and BergstroÈm 1989). How- 1987; Gill 1989; Neiland and Wilcock 1995). In six inves- ever, no behavioral tests were performed to show whether tigated species of deceptive orchids, 75±98% of all plants these variations of odor bouquets from various ¯owers of the lacked fruit set (Calvo 1990). In a population of O. sphegodes same or of different plants can be learned by pollinators. near Illmitz (Austria), only 4.9% of 887 plants were visited These important aspects of the behavior of pollinators at by a pollinator, as revealed by either pollinia removal, pol- deceptive ¯owers thus remain to be documented. lination, or both (Ayasse et al. 1997). A high specialization The aim of this study was to determine whether individual in Ophrys orchids to one or a very few pollinators may be ¯ower-speci®c olfactory recognition signals exist in the or- the most important reason for the observed low pollination chid O. sphegodes and if the pollinator males learn and rec- frequency. As an adaptation to this, ¯owers are relatively ognize odors of individual ¯owers. We addressed these goals long-lived and the opportunity for pollination is maximally through: (1) behavioral learning experiments in the ®eld with extended (Dafni and Bernhardt 1990). Furthermore, polli- pollinating male bees of Andrena nigroaenea; (2) quantitative nated ¯owers produce a high number of seeds (van der Cingel chemical analyses of the odors from individual ¯owers; (3) 1995). gas chromatography with electroantennographic detection Such a low pollination frequency should lead to the evo- (GC-EAD) analyses to identify biologically active com- lution of strategies that increase the chance that males visit pounds (compounds perceived by the males); and (4) behav- more than one ¯ower on the same plant, increasing the num- ioral experiments in the ®eld to show how self-pollination is ber of pollination events and therefore the number of seeds avoided when a pollinator visits more than one ¯ower on the produced. However, pollinators successively visiting several same in¯orescence. ¯owers within an in¯orescence may increase the frequency of geitonogamy (pollination of a neighboring ¯ower). Fur- MATERIALS AND METHODS thermore, if they return to an already visited ¯ower, autogamy Study Species (self-pollination) may occur. Although autogamy or geiton- ogamy have an obvious advantage in maintaining a high seed Ophrys sphegodes sphegodes Miller grows on alkaline, dry production even if pollinators are scarce, inbreeding depres- to damp soils in meadows, grassland, and scrubland of the sion may occur. In Orchis morio, self-pollinated ¯owers Mediterranean region and Central Europe (Delforge 1994). showed a delimited fruit formation compared to outcrossed In Austria, O. sphegodes is mostly con®ned to the east and ¯owers (Neiland and Wilcock 1995), and in Caladenia ten- southeast. It ¯owers in April and May. Flower spikes are taculata outcrossing seeds developed more quickly than rarely more than 20 cm tall and one in¯orescence is produced selfed seeds (Peakall and Beatie 1996). Due to these effects per stem. Most plants bear two to ®ve ¯owers that open on ®tness, pollinators should visit several different plants in successively. Flowers are long-lived (14±21 days) if unpol- a population and selection should enhance reproductive suc- linated, fading rapidly after pollination (Schiestl et al. 1997). cess at the level of population as well as the individual plant. Ophrys sphegodes is pollinated by A. nigroaenea males (God- Another reason for the low frequency of pollination in fery 1925; Paulus and Gack 1986, 1990), a solitary bee spe- sexually deceptive orchids may be the learning capabilities cies that is common in the area of our study and nests in the of pollinators. Behavioral experiments have shown that male soil (Westrich 1989). Because females' nests are mostly dis- bees are able to learn the distinctive odor bouquets of indi- persed, males search for and mate with females at non-re- vidual female bees during mating attempts (Smith and Ayasse source-based landmarks or patrol along resources visited by 1987; Dutzler and Ayasse 1996) and that they use this in- females for self-nourishment (¯owers). Sex pheromones formation to avoid females they have already mated with. emitted by females are important cues used by males to locate Bees might respond in a similar manner to Ophrys fragrances and/or recognize mates at these rendezvous sites (TengoÈ and individual unpollinated ¯owers might vary in their odor 1979). Typically, males engage in a scramble contest for bouquets both within populations and even within the same virgin females. in¯orescence to increase the frequency of pollinated ¯owers. The evolutionary mechanism could be frequency-dependent Study Sites response of male pollinators to ¯oral signals (Paulus 1988). We collected plants of O. sphegodes at two distinct pop- This would promote selection for ¯oral scent variation. How- ulations: (1) the Bisamberg population, approximately 10 km ever, in sexually deceptive orchids, it has not yet been de- to the north of Vienna; and (2) the Illmitz population, close termined whether different odors are produced by the ¯owers to the eastern border of Lake Neusiedl. The distance between of different plants or by the ¯owers within a single in¯ores- the populations is 30 km. All behavioral tests were conduced cence. in Oberweiden and Weikersdorf (Eastern Austria). To avoid The hypothesis that variation in ¯ower characteristics in habituation of the males, we used several test locations (10± non-food-rewarding ¯owers inhibits the formation of a search 100 m apart) that were changed after a few tests. image by the pollinator (Heinrich 1975; Nilsson 1980; Ack- erman 1986) and interferes with its associative learning pro- Flower Odor Sampling cess was tested by Moya and Ackerman (1993), who found high levels of ¯ower fragrance variation in the nectarless, Individual Ophrys labella were cut from ¯owers and ex- moth-pollinated orchid Epidendrum ciliare. Intraspeci®c var- tracted in 0.5-ml pentane (Uvasol, Merck, Darmstadt, Ger- iation in ¯oral fragrances has also been documented for nec- many) at 20ЊC for 2 days. The labella were then removed tar-producing orchids, including Platanthera stricta (Patt et and the samples were stored at Ϫ20ЊC until chemical analyses REPRODUCTIVE STRATEGIES IN OPHRYS SPHEGODES 1997 were performed. Because we could show that extracts of cut The resulting principal components (PCs) with an eigenvalue labella as well as mixtures of synthetic compounds mixed above one were used to test for differences in odor bouquets according to labella extracts modulate pollinator behavior by means of a discriminant function analyses (DFA). We (see Schiestl et al. 1999; this study), we can exclude effects compared the uppermost ¯owers of different plants growing of cutting off labella. Therefore, the results of our chemical in different populations and the uppermost and the second analyses re¯ect the volatiles released by intact ¯owers. uppermost ¯owers of an in¯orescence using the calculated factor scores for each case on the principal axes (SPSS 1997). Chemical Analyses The standardized discriminant function coef®cients and the factor loadings after varimax rotation were used to assess the Labella solvent extracts were analyzed on a gas chro- importance of individual compounds. We considered a com- matograph HP 5970 (Hewlett-Packard, Palo Alto, CA) pound to have a high factor loading if the loading was above equipped with a DB5 capillary column (30 m ϫ 0.32 mm 0.5. internal diameter). The gas chromatograph was operated To compare the distances (dissimilarities) of odor bouquets splitless at 120ЊC for 30 sec, followed by a programmed produced by various ¯owers, we used the factor scores. Val- increase to 280ЊCat4ЊC/min. For quantitative analysis, n- ues for factor scores were ®rst standardized to Z-scores (with octadecane was added as an internal standard. Structure elu- a mean of zero and a standard deviation of one), then a cidation of individual compounds was based on gas chro- dissimilarity matrix was computed by measuring the Euclid- matography±mass spectrometry (GC-MS) analysis (VG70/ ean distance values between cases. Measurements were based 250 SE instrument, Vacuum Generators, Manchester, Eng- on the factor scores of two PCAs performed either with the land, linked to a HP 5890; gas chromatographic condition as relative proportions of 24 GC-EAD active or with 71 inactive mentioned above); mass spectra were compared with those compounds. Because the measurements of the active and in- reported in the literature (Mc Lafferty and Stauffer 1989) and active compounds were based on a different number of var- gas chromatographic retention times (coinjection) with those iables, the resulting Euclidean distance values had a different of authentic reference samples. Double-bond positions in un- range. To make the data of the dissimilarity matrices of the saturated compounds were assigned according to Buser et al. active and nonactive compounds comparable, we rescaled the (1983) and Dunkelblum et al. (1985). The stereochemistry distance values to a range of zero to one. of double bonds was determined by comparison of retention times using corresponding reference samples, including Behavioral Tests DMDS derivatives. The erythro- and threo-aducts can be well separated by GC. On the same day the behavioral experiments were per- formed, ¯owers were collected at random from the Bisamberg Electrophysiology and Illmitz populations. Individual ¯owers were ®xed on in- sect pins and offered to the males at separate locations. Cop- To ®nd chemical compounds within Ophrys odors to which ulation attempts were recorded during a 3-min test period. male antennal olfactory receptor neurons are sensitive, the Between 10 to 20 males from each location (two locations tip of an excised male antenna was cut off and the antenna in Weikersdorf and six locations in Oberweiden) were in- was mounted between two pipet electrodes containing insect- dividually marked with enamel paint on the top of the thorax. Ringer, and grounded via an Ag-AgCl wire. The tip electrode Observations of individually marked males indicate that was connected to a high-impedance DC ampli®er with au- males do not move freely throughout the entire mating area. tomatic baseline drift compensation. A Hewlett-Packard 5890 Most marked males could be observed at spots no more then gas chromatograph equipped with a DB-Wax capillary col- 10 m away from where they were marked. Therefore, mostly umn (30 m ϫ 0.32 mm internal diameter) and ef¯uent split the same males were tested within any one area. was used for simultaneously recording ¯ame ionization (FID) and electroantennographic (EAD) signals. Helium was used Learning Experiments with the Pollinating Male Bees as carrier gas and the ef¯uent split ratio was approximately 1:1. The gas chromatograph was operated splitless at 50ЊC To test the hypothesis that males can learn the distinctive for 1 min, followed by opening the split and programming odor bouquet of individual ¯owers, in a ®rst behavioral ex- to 230ЊCat10ЊC/min. The outlet for the EAD was placed in periment a ¯ower was tested in one location and retested 2 a charcoal-®ltered, moisturized airstream ¯owing over the min afterwards in the same location. After 2 min, the same antennal preparation at a speed of 0.5 m/sec. One microliter ¯ower was tested at another location of mostly other males per sample was injected into the column. GC-EAD active patrolling. Choices of the two locations were random for each compounds were identi®ed by GC-MS (methods, see above) test. While performing the tests, we determined that two of and synthesized. Additional GC-EAD analyses with these 20 males we marked in one of the test locations could also compounds were performed. be found in the neighboring test location. Because males that moved between both locations had been able to learn a tested ¯ower offered at a second test location, we excluded the Statistical Analysis results of those tests. Three principal component analyses (PCAs) were per- A second experiment was performed to exclude both visual formed on a set of 114 labella solvent extracts. We used the learning of a ¯ower and odor marking by a pollinating male. relative proportions of 24 GC-EAD active components for Males were offered a ¯ower and a dead, odorless female of two PCAs and of 71 nonactive components for a third PCA. A. nigroaenea. These dummies had been Soxhlet-extracted 1998 MANFRED AYASSE ET AL.

TABLE 1. Comparison of the mean absolute amounts (␮g) of GC- Behavioral Experiments to Prove How Self-Pollination Is EAD active aldehydes and saturated esters in labellum extracts of the Prevented uppermost (n ϭ 20) and the second uppermost ¯owers (n ϭ 20) of an in¯orescence. Ten plants with three ¯owers each were picked on the day of the behavioral experiments by sampling at random from Flower stem position the Bisamberg population. Plants were kept in glass tubes Uppermost Second uppermost Differ- ®lled with water and transported to Oberweiden. Single plants Compounds Mean Ϯ SEM Mean Ϯ SEM ence were offered to male bees at two separate locations 30 m Octadecanal* 0.14 Ϯ 0.02 0.24 Ϯ 0.02 0.10 apart. The number of individually marked males that visited Eicosanal* 0.40 Ϯ 0.05 0.59 Ϯ 0.05 0.19 2-Nonyl tetradecanoate 0.84 Ϯ 0.07 1.02 Ϯ 0.12 0.18 one ¯ower of an in¯orescence and returned to visit another 2-Nonyl hexadecanoate*1 ¯ower of the same in¯orescence were recorded. We compared ϩ Dodecyldodecanoate 0.80 Ϯ 0.09 1.34 Ϯ 0.16 0.54 the time spent on any visited ¯ower and the lapse between Dodecyl tetradecanoate* 0.41 Ϯ 0.04 0.94 Ϯ 0.12 0.53 leaving a visited ¯ower of an in¯orescence and arriving on * Signi®cant difference, t-test, P Ͻ 0.001. 1Compounds that could not be another ¯ower of the same in¯orescence (bee ¯ying time). separated by gas chromatography. In all Ophrys species, the pollen is packed in one mass, the pollinarium (van der Cingel 1995). During visits by pol- linators, the pollinia become attached to the insect's head or tip of the abdomen, and a bending mechanism subsequently in dichlormethane for 48 h, dried and ®xed on an insect pin. takes place, whereby the pollinium swings forward through The ¯ower was upside down and under the dummy bee, where an angle of about 90Њ. After this, the pollinium is in the it was out of reach of the incoming male. Therefore, males correct position to ®t into the stigma of another ¯ower (Dar- stimulated by the scent of a ¯ower touched the dummy bee win 1862). This bending, which takes place within a matter only. After 2 min, the same dummy bee was retested with a of seconds or several hours, decreases geitonogamy and en- second ¯ower of another plant. Our working hypothesis was hances the likelihood of cross-fertilization. To do so, the time that either marking a visited dummy with repellents (anti- required for the pollen stalk to complete its bend should be aphrodisiac) or learning the visual cues of the two ¯owers longer than the amount of time that a pollinator spent on a should result in a decreased attractiveness during the retest plant (Catling and Catling 1991). In O. sphegodes, we re- of a dummy female. corded the time needed by a removed pollinium to complete To test if males can discriminate between odor bouquets its bend. in ¯owers of the same or different plants, a ¯ower was tested in one location, and 2 min afterwards either a ¯ower of the RESULTS same in¯orescence or a ¯ower of another plant of the same population or of a plant of a second population was tested Chemical Analyses in the same location. We identi®ed 106 compounds in solvent extracts of O. Another experiment was performed to test the hypothesis sphegodes labella. The volatile bouquets are dominated by that males in successive visits of an in¯orescence differen- saturated nonterpenoic esters, n-alkanes, n-alkenes, and n- tiate between ¯owers by learning the ¯ower-speci®c patterns alkadienes with chain lengths of 19±33 carbons and alde- of chemical compounds. Among the compounds that were hydes with chain lengths of 9±22 carbons. In addition, small found to be GC-EAD active, esters and aldehydes did not amounts of fatty acids and terpenoic esters were identi®ed elicit mating behavior in the male pollinators (F. P. Schiestl, (see Fig. 1). unpubl. data). Our chemical analyses showed the differences between the uppermost and the second uppermost ¯owers of Electrophysiology an in¯orescence were mainly in the concentrations of these The results of simultaneous FID and EAD analyses per- compounds. Therefore, our hypothesis that the males learn formed with labellum extracts of O. sphegodes showed that ¯ower-speci®c patterns of esters and aldehydes was tested n-alkanes and n-alkenes, aldehydes, saturated nonterpenoic by offering the uppermost ¯ower of an in¯orescence to the esters, E,E-farnesyl hexanoate, E,E-farnesol, and nonanoic males in one location. Thereafter, we impregnated the la- acid triggered receptor potentials in the antennae of males bellum surface with 20 ␮l of a mixture of synthetic esters (see Fig. 2). From 24 compounds that were found to be bi- and aldehydes. The mixture contained only those compounds ologically active, 14 could be identi®ed as n-alkanes or n- whose total amount of odor in the labellum extracts was alkenes with the double bond in position 12, 11, or 9. Re- higher in the second uppermost then in the uppermost ¯owers cently, the importance of these compounds in stimulating of an in¯orescence (Table 1). From each compound, we added mating behavior in the males was shown by behavioral ex- exactly the differential amount between both groups of ¯ow- periments (Schiestl et al. 1999). ers. Because 2-nonyl hexadecanoate and dodecyl dodeca- noate could not be separated by GC analyses, our test mixture Intraspeci®c Variation of the Odor Bouquets of Labellum contained 50% of both compounds. The impregnated ¯owers Extracts were tested in the same location with the same males pa- The ®rst PCA included the relative proportions of 24 GC- trolling. In a control experiment, ¯owers were retested after EAD active compounds in 82 labellum extracts. We used impregnation with solvent only (20 ␮l pentane Uvasol). only samples of the uppermost ¯owers from plants of the REPRODUCTIVE STRATEGIES IN OPHRYS SPHEGODES 1999 labellum, with over 100 compounds identi®ed (x, compounds not yet identi®ed). Ophrys sphegodes . 1. Gas chromatogram of the solvent extract of an IG F 2000 MANFRED AYASSE ET AL.

FIG. 2. Simultaneous ¯ame ionization/detection and electroantennographic analyses performed with an Ophrys sphegodes labellum extract. Mostly saturated and unsaturated hydrocarbons triggered receptor potentials in the antennae of males (*, compounds that could not be separated with the gas chromatography parameters used).

two populations (Bisamberg and Illmitz). Six PCs with an compounds from different chemical classes of compounds to eigenvalue above one explained 73.4% of the total matrix be associated with each of the PCs. variation. In most cases, compounds from the same chemical classes were found to be associated with each of the PCs. Comparison of Odor Bouquets from Flowers of Different A further PCA was based on the relative proportions of Populations the GC-EAD active compounds of 40 labellum extracts from The odor bouquets comparisons based on the relative pro- uppermost and second uppermost ¯owers collected in Illmitz. portions of 24 GC-EAD active compounds showed signi®cant Six PCs with an eigenvalue above one explained 79.9% of differences between populations. The labellum extracts of the the total matrix variation. As in the ®rst PCA, predominantly uppermost ¯owers from Bisamberg (n ϭ 33) and Illmitz (n compounds from the same chemical classes were found to ϭ 49) were well separated by a DFA performed with the be associated with each of the PCs. calculated factor scores of six PCs (␹2 ϭ 47.3, df ϭ 6, P Ͻ The third PCA was performed with the relative proportions 0.001). The standardized canonical discriminant function co- of the nonactive compounds from the same labellum extracts ef®cients and the factor loadings showed that the differences as the second PCA. It revealed highly variable odor bouquets between the populations were mainly in concentrations of n- produced by various ¯owers. Fifteen PCs with an eigenvalue alkanes, nonacosene (double-bond positions 11 and 12), (Z)- above one explained 88.9% of the total matrix variance. The 9-tricosene, as well as aldehydes (octadecanal, eicosanal), results of the PCA using varimax rotation showed various E,E-farnesol, and nonanoic acid. REPRODUCTIVE STRATEGIES IN OPHRYS SPHEGODES 2001

FIG. 3. Comparison of the relative proportions (mean ϩ SEM) of the main chemical classes of compounds in labellum extracts of the uppermost ¯owers from Bisamberg (n ϭ 33) and Illmitz (n ϭ 49; FIG. 4. Comparison of the relative proportions (mean ϩ SEM) of **, signi®cant difference, t-test, P Ͻ 0.01). the main chemical classes of compounds in labellum extracts of the uppermost and the second uppermost ¯owers of an in¯orescence (**, signi®cant difference, t-test, P Ͻ 0.01). A univariate comparison of the relative proportions of the main chemical classes of compounds is shown in Figure 3. n-alkanes and aldehydes were relatively more abundant at by (uppermost) ¯owers of different plants showed signi®- Illmitz, n-alkenes were higher at Bisamberg, and nonterpen- cantly larger distance values (mean Ϯ SEM ϭ 0.39 Ϯ 0.02, oic esters did not differ between populations. n ϭ 70; t ϭϪ4.49, df ϭ 89, P Ͻ 0.001). When we used the relative proportions of GC-EAD nonactive compounds, the Comparison of the Odor Bouquets Produced by Flowers of calculated mean Euclidean distance values were signi®cantly Different Stem Positions larger as compared to those based on GC-EAD active com- The multivariate comparison performed with the factor pounds (Fig. 5). scores of six PCs showed a signi®cant difference between the uppermost and the second uppermost ¯owers of an in- Learning Experiments with the Male Pollinators ¯orescence (DFA: ␹2 ϭ 35.1, df ϭ 6, P Ͻ 0.001). Because Field learning experiments showed that O. sphegodes ¯ow- the standardized canonical discriminant function coef®cients ers ®xed on insect pins were highly attractive to A. nigroaenea were highest for PC1 and PC2, nonterpenoic esters, aldehydes males, which tried to copulate with the ¯owers. A retest of (octadecanal, eicosanal), E,E-farnesol, nonanoic acid, and the same ¯ower in the same location triggered signi®cantly (Z)-9-heptacosene contributed most to the differences be- fewer copulation attempts and pouncing events as compared tween ¯ower positions. When comparing the relative amounts to the ®rst test (t ϭ 2.64, df ϭ 22, P ϭ 0.02; Sokal and Rohlf of the main chemical classes of compounds, we found that 1995; see Fig. 6, left). When a ¯ower was transferred to a n-alkenes were relatively more abundant in the uppermost second test location where different males patrolled, its at- ¯owers, nonterpenoic esters and aldehydes were higher in the second uppermost ¯owers, and n-alkanes did not differ be- tween ¯owers of different stem positions (Fig. 4). There was no signi®cant difference in the total amount of odor in the labellum extracts from the uppermost (mean Ϯ SEM ϭ 11.59 Ϯ 0.58 ␮g, n ϭ 20) and second uppermost ¯owers (mean Ϯ SEM ϭ 12.90 Ϯ 0.74 ␮g, n ϭ 20) of an in¯orescence (t-test, t ϭϪ1.39, df ϭ 38, P ϭ 0.174). A comparison of the amount of odor of single compounds, however, showed an increase in certain aldehydes and saturated esters (Table 1).

Similarities of Odor Bouquets Produced by Individual Flowers We found the smallest mean Euclidean distance value when we compared the relative proportions of GC-EAD active compounds from the uppermost and second uppermost ¯ower of the same in¯orescence (mean Ϯ SEM ϭ 0.21 Ϯ 0.03, n FIG. 5. Comparison of Euclidean distance values (mean Ϯ SEM) ϭ 20, see Fig. 5), indicating a high similarity of odor bou- of odor bouquets from ¯owers within and between plants (different quets. Within the Illmitz population, odor bouquets produced letters indicate signi®cant differences; t-test, P Ͻ 0.05). 2002 MANFRED AYASSE ET AL.

FIG. 6. Field studies on the learning of distinct odor bouquets from individual ¯owers by male Andrena nigroaenea. Flower attractiveness measured by the mean number of copulation attempts (mean Ϯ SEM) that males performed on a ¯ower. Left: A ¯ower was offered in location A for 3 min. It was retested in the same location and in a second location (B) where other males patrolled. Right: Experiment to exclude both visual learning of a ¯ower and odor marking by a pollinating male. Males were offered a dead, odorless female A. nigroaenea and a ¯ower ®xed on an insect pin upside down beneath the dummy bee, invisible and out of reach of the incoming male. After 2 min, the same dummy bee was retested with a ¯ower of another plant (different letters indicate signi®cant differences; t-test, P Ͻ 0.05). tractiveness increased again (t ϭ 1.89, df ϭ 22, P ϭ 0.07) Behavioral Observations and Experiments to Prove How and was not signi®cantly different to the attractiveness in the Self-Pollination Is Prevented ®rst test (t ϭ 0.54, df ϭ 22, P ϭ 0.59). To exclude both From 30 males that visited one ¯ower of an in¯orescence, visual learning of a ¯ower and odor marking by a pollinating 20 males (67%) returned to visit a second ¯ower of the same male, we conducted retests on dead, odorless females in the in¯orescence. Some males even visited a third ¯ower of an same location using the same males that were stimulated to in¯orescence. The time from the beginning until the end of mate by the scent of a second ¯ower of another plant. These the bending mechanism, which amounts to the time needed experiments resulted in the same or even slightly higher at- by a removed pollinium to bend into a correct position for tractiveness of the dead female (t ϭϪ1.08, df ϭ 36, P ϭ ®tting into the stigma of another ¯ower, is longer than the 0.29; see Fig. 6, right). total time for visiting the ®rst and second ¯ower of an in- A retest in the same location using a second ¯ower either ¯orescence, including the transit time between ¯owers (Fig. from the same in¯orescence or from another plant from the 7). same or a different population showed that the attractiveness of the two ¯owers was the same (Table 2). DISCUSSION We tested the hypothesis that the males learn ¯ower-spe- ci®c patterns of esters and aldehydes by ®rst offering the Intraspeci®c Variation of Flower Scent uppermost ¯ower of an in¯orescence to males in one location Orchids of the genus Ophrys produce a huge number of and then by retesting the same ¯ower after impregnation with volatiles (Borg-Karlson 1990). By GC-MS analyses and coin- esters and aldehydes. In a retest the attractiveness of the jection technique we identi®ed a complex mixture of com- ¯ower was not diminished (Table 3). However, when the pounds in the labellum extracts of O. sphegodes. However, ¯ower in the retest was impregnated with solvent only (con- behavioral experiments and EAD analyses have shown that trol), it elicited signi®cantly fewer copulation attempts than only some of these compounds are active in stimulating mat- in the ®rst test (Table 3).

TABLE 3. Learning experiments to test hypothesis that males learn TABLE 2. Field studies of learning of individual ¯owers' distinct odor individual ¯owers of an in¯orescence by the patterns of aldehydes and bouquets by male Andrena nigroaenea. In all test groups, the attrac- esters. The uppermost ¯ower of an in¯orescence was offered in a tiveness (mean number of copulation attempts) of both ¯owers tested location for 3 min and retested after impregnation with (1) aldehydes one after another in the same location was the same. and esters; or (2) pentane (control). Attractiveness was measured by the mean number of copulation attempts. Attractiveness Attractiveness First ¯ower Second ¯ower t-test Test groups (n) Mean Ϯ SEM Mean Ϯ SEM P First test Second test t-test Test groups (n) Mean Ϯ SEM Mean Ϯ SEM P Flowers of the same in¯orescence (24) 8.83 Ϯ 1.59 10.33 Ϯ 1.44 0.49 Flowers impregnated with Flowers of different aldehydes and saturated plants (31) 7.68 Ϯ 1.47 8.87 Ϯ 1.72 0.60 esters (15) 8.33 Ϯ 1.71 8.47 Ϯ 1.49 0.95 Flowers of different Flowers impregnated with populations (20) 14.65 Ϯ 2.55 16.70 Ϯ 2.46 0.57 pentane (10) 7.00 Ϯ 0.92 4.00 Ϯ 0.60 0.01 REPRODUCTIVE STRATEGIES IN OPHRYS SPHEGODES 2003

tracting volatiles of Ophrys ¯owers should also show a fairly low degree of variability to match the signal that males are able to perceive. Floral scents can in¯uence pollinators in several ways: they attract from a distance, trigger searching behavior, and act as cues for alighting or probing by the pollinators (Faegri and van der Pijl 1979). In both rewarding and sexually de- ceptive orchids, variability of the scent of individual ¯owers may in¯uence the frequency of visits and thereby the overall ®tness of individual plants under conditions of a low polli- nator visitation rate (Moya and Ackerman 1993). Our hy- pothesis that O. sphegodes ¯owers from different plants and even from the same in¯orescence produce different polli-

FIG. 7. Behavioral experiments to test how self-pollination is pre- nator-attracting olfactory recognition signals was con®rmed vented. We recorded and compared the time (median) spent on any by the results of our chemical analyses. The ¯oral volatile visited ¯ower, bee ¯ying time between visiting ¯owers of the same bouquets differed between as well as within populations. in¯orescence, and time needed by a removed pollinium to bend into Based on the statistical analyses, mainly n-alkanes and n- a correct position for ®tting into the stigma of another ¯ower. Boxes represent interquartile ranges and are divided at the median. Stems alkenes showed a characteristic population-speci®c odor pat- represent smallest or largest observations within 1.5 interquartile tern, whereas esters varied only in dependence of the stem ranges of the bottom or the top of the boxes. position of a ¯ower. Furthermore, our data showed for the ®rst time intraspeci®c ¯oral scent variation of ``behaviorally active compounds'' in a sexually deceptive orchid. ing behavior in males (Schiestl et al. 1999). Similar infor- Differences in the scent of three forms of O. insectifera, mation also exists for other Ophrys species (Priesner 1973; another sexually deceptive orchid, have been described by Kullenberg and BergstroÈm 1976; Borg-Karlson 1990). In O. Borg-Karlson et al. (1993). However, because all three forms lutea, a blend of synthetic compounds consisting of aliphatic attracted males of different pollinator species, the forms may 1-alcohols, 2-alcohols, and terpenes elicited the same level represent separate species (Paulus and Gack 1990; van der of excitation in Andrena fuscipes males as did whole labellum Cingel 1995). Considerable quantitative and qualitative scent extracts (Borg-Karlson and TengoÈ 1986). Why do Ophrys variations also exist within and among plants of food-de- ¯owers produce compounds that are not behaviorally active? ceptive orchids (Tollsten and BergstroÈm 1989; Knudsen and With the exception of n-alkenes, all other compounds we Tollsten 1993; Moya and Ackerman 1993). In Epidendrum identi®ed in O. sphegodes labellum extracts are abundant in ciliare, ¯oral fragrances vary according to the age of ¯owers, cuticular waxes of plants and are said to have the primary their stem positions, and their genotypes (Moya and Ack- function to protect the plant from dehydration (Eigenbrode erman 1993). Scents produced by individual O. sphegodes and Espelie 1995). Many of the compounds we identi®ed in ¯owers may vary due to that plant individual's genotypic O. sphegodes labellum extracts can also be found in other makeup or within an in¯orescence as a consequence of in- Ophrys species (Borg-Karlson 1990) and could therefore have teractions of the genotype with the environment. Post-pol- such a function. Compounds that are not used to attract pol- lination changes in the production of scent were shown to linators may furthermore act as a solvent for the male-at- guide pollinators of O. sphegodes to the unpollinated ¯owers tracting volatiles or they may repell other bee species of an in¯orescence (Schiestl et al. 1997), thereby increasing (Francke 1986). Another selective pressure for the production the number of pollinated ¯owers. The changes involved a of a wide array of compounds in Ophrys orchids might have decrease of the total amount of scent produced as well as an been a frequent adaptation to new pollinator species. Over alteration of the odor bouquet. E, E-farnesyl hexanoate is the longer terms, it may be a more economic evolutionary strat- major component of the Dufour's gland secretion in breeding egy to use a broad spectrum of components and to alter the (mated) females of the pollinator bee, A. nigroaenea, and is pattern of compounds that already exist than to synthesize produced by O. sphegodes ¯owers after pollination. Behav- new compounds that may involve major changes in the bio- ioral tests showed that E, E-farnesyl hexanoate has a key synthetic pathways. Which of these reasons is valid in ex- function as a repellent signal of pollinated ¯owers (Schiestl plaining the large number of compounds in Ophrys orchids and Ayasse 2001). remains unanswered. In O. sphegodes, we found a smaller intraspeci®c variation Does Scent Variation In¯uence the Behavior of the of the behaviorally active compounds as compared to the Pollinators? nonactive compounds. Why do behaviorally active com- pounds show a smaller ¯ower-speci®c variation? One expla- Our behavioral tests indicate that male bees learn the odor nation for the smaller intraspeci®c variation of the behavior- bouquets of individual ¯owers during mating attempts in the modulating compounds may be a higher selective pressure same manner as when interacting with their own females on the pollinator-attracting communication signal. Sex pher- (Smith and Ayasse 1987; Smith 1993; Dutzler and Ayasse omones of female insects consist of highly species-speci®c 1996). Because females of most bee species are monandrous mixtures of compounds with a fairly low degree of variability and mate soon after emergence (Eickwort and Ginsberg (Baker 1989; LoÈfstedt et al. 1991). Therefore, the male-at- 1980), selection should proximately reward males capable of 2004 MANFRED AYASSE ET AL. precisely recognizing promising mates, which should ulti- males, along with an identical or similar scent of the second mately lead to a mating behavior wherein males spend no ¯ower offered, should have resulted in a diminished attrac- time chasing unreceptive females (Ayasse et al. 1999). How- tiveness. ever, the diminished attractiveness of previously visited ¯ow- In O. sphegodes, previous studies showed that hydrocar- ers in subsequent encounters can be interpreted with several bons are responsible for eliciting male copulatory behavior hypotheses: (1) males learn the location where they visited (Schiestl et al. 1999), whereas this was not achieved by syn- a nonrewarding ¯ower; (2) males mark ¯owers with repel- thetic blends of esters and aldehydes (F. P. Schiestl, unpubl. lents (``antiaphrodisiac,'' Kukuk 1985) that deter other males data). Here we demonstrate that, by varying the relative pro- from copulatory attempts; (3) males learn visual cues of vis- portions of saturated esters and aldehydes between ¯owers ited ¯owers; (4) ¯ower odor dissipates during the ®rst test of different stem positions, a plant may take advantage of of a ¯ower resulting in a diminished attractiveness of a re- the learning abilities of the pollinators and in¯uence ¯ower tested ¯ower; or (5) males learn that ¯owers do not copulate visitation behavior. Contrary to the idea that habituation of and during subsequent encounters they do not respond in the males to the scent of a visited Ophrys ¯ower would moderate same intensity to those learned odors (motivational decay). their response to all ¯owers of an in¯orescence (Paulus 1988; The learning of precise locations was shown in wasps Nilsson 1992), we found in O. sphegodes that 67% of the (Evans 1966). However, our ®nding that a retest in the same males that visited one ¯ower of an in¯orescence returned to location with new ¯owers did not diminish attractiveness visit a second ¯ower of the same in¯orescence. Data on the (Table 2) argues against this hypothesis and also against mo- frequency of repeated visits to various ¯owers of an in¯o- tivational decay. Our data indicate that males do not learn rescence by a single pollinator are rare in sex-deceptive or- ¯owers by visual cues and furthermore do not mark ¯owers chids. In the wasp-pollinated orchid Drakea glyptodon, most with repellents (antiaphrodisiacs) that deter other males from pollinator males immediately leave the area after they visit copulation attempts: Because the same dead, odorless female a ¯ower and will not visit nearby ¯owers (Peakall 1990). was retested in the same location (with the same males that Such behavior produces long-distance pollen ¯ow and out- were stimulated to mate by the scent of an Ophrys ¯ower), crossing. However, plants in D. glyptodon normally bear one marking of the female with a repellent or learning by visual ¯ower only. Therefore, revisiting a plant may cause ®tness cues should have decreased the attractiveness during the re- disadvantage by self-pollination and does not increase re- test of the female (Fig. 6, right). Visual cues might play a productive success in the same range as in O. sphegodes. more important role in other Ophrys species, in which optical Pollinator movement between ¯owers on an in¯orescence cues are more pronounced or where the overall similarity of may also lead to geitonogamous pollination, like in the sex- the Ophrys ¯ower to the pollinator female is obvious like in ually deceptive orchid Prasophyllum ®mbria, where geiton- O. vernixia (Paulus and Gack 1994). When a tested ¯ower ogamous transfer of pollen was found in 22% of all polli- was transferred to a new test location with other patrolling nations (Peakall 1989). The genetic consequence of self-pol- males, its attractiveness increased again (Fig. 6, left). This lination is inbreeding depression (Proctor et al. 1996). There- leads us to exclude ¯ower odor dissipation as a reason for fore, in orchids mechanisms have evolved to prevent the diminished attractiveness of ¯owers tested a second time autogamy and geitonogamy (van der Cingel 1995). in the same test location. Our data clearly show that experience with a ¯ower odor How is Autogamy and Geitonogamy Prevented in Ophrys decreases the ¯ower's attractiveness for pollinators and that sphegodes? this habituation is based on the learning of chemically distinct In O. sphegodes, we found that the time needed by a re- odor signals in individual ¯owers. In the Australian sexually moved pollinium to bend into a correct position to ®t into deceptive orchid genera Chiloglottis, Caladenia, and Drakea, the stigma of another ¯ower is longer than the total time for behavioral experiments with male pollinators (thynnine males to visit the ®rst, second, and third ¯ower of an in¯o- wasps) showed comparable results. Following an initial en- rescence and in addition to move between these ¯owers. The counter, males avoided the orchid ¯owers (Peakall 1990; likelihood that geitonogamy takes place is therefore reduced Peakall and Handel 1993; Peakall and Beattie 1996). Peakall in O. sphegodes. Similar data exist for other orchids. In Pla- (1990) discussed a ``site-speci®c refractory response'' of the tantera blephariglottis, pollinators visited three to four ¯ow- males or alternatively learning of a recognition pheromone. ers per in¯orescence and the average time spent on an in¯o- Because conclusive experiments with changed labella tested rescence was 34 sec, whereas complete bending of the pollen in different locations were not performed, however, he could stalk requires about 1 min (Cole and Firmage 1984). not rule out one of both explanations. Evolution of Reproductive Strategies in Ophrys sphegodes Does Flower-Speci®c Scent Variation Lead Pollinators to According to Proctor et al. (1996) Ophrys orchids have Revisit a Plant? evolved ``one of the most remarkable pollination mechanisms Our behavioral experiments where two O. sphegodes ¯ow- found in any plants.'' A strong specialization of O. sphegodes ers were offered successively in the same male location in- in attracting only males of one pollinator species plays an dicate that individual ¯owers from different plants or within important role in preventing hybridization and pollen loss. an in¯orescence smell differently. These experiments are As a consequence of this and of the highly sophisticated therefore consistent with our chemical analyses. Learning the learning capabilities of bees (Smith and Ayasse 1987; Smith scent of the ®rst ¯ower of an in¯orescence offered to the 1993), however, the pollinator visiting rate is low (Ayasse REPRODUCTIVE STRATEGIES IN OPHRYS SPHEGODES 2005 et al. 1997), as in other deceit-pollinated orchids (Fritz and Ayasse, M., W. Engels, G. LuÈbke, T. Taghizadeh, and W. Francke. Nilsson 1996), resulting in a low number of fruits produced. 1999. Mating expenditures reduced via female sex pheromone modulation in the primitively eusocial halictine bee, Lasioglos- Mechanisms have evolved in O. sphegodes to increase re- sum (Evylaeus) malachurum (Hymenoptera: Halictidae). Behav. productive success. Characteristic variation in the relative Ecol. Sociobiol. 45:95±106. proportions of the chemical compounds of ¯owers within an Baker, T. C. 1989. Sex pheromone communication in the Lepidop- in¯orescence uses the learning abilities of the male polli- tera: new research progress. Experientia 45:248±262. Borg-Karlson, A.-K. 1987. Chemical basis for the relationship be- nators, raising the chance of more than one ¯ower being tween Ophrys orchids and their pollinators. III. Volatile com- visited by the same male bee. Assuming that a ¯ower-visiting pounds of species in the Ophrys sections Fuci¯orae and Bom- pollinator is carrying pollinia packed up from ¯owers of an- byli¯orae as insect mimetic attractants/excitants. Chemica Scrip- other plant and would visit at least two ¯owers on an in¯o- ta 27:313±325. rescence, male and female reproductive success of a plant ÐÐÐ. 1990. Chemical and ethological studies of pollination in the genus Ophrys (Orchidaceae). Phytochemistry 29(5): would increase. First, the pollinator could remove pollinia of 1359±1387. the visited ¯owers and carry them to the ¯owers of another Borg-Karlson, A.-K., and J. TengoÈ. 1986. Odour mimetism? Key plant; second, at least two ¯owers of the visited in¯orescence substances in the Ophrys lutea-Andrena pollination relationship could receive pollen and produce seeds. (Orchidaceae-Andrenidae). J. Chem. Ecol. 12:1927±1941. Borg-Karlson, A.-K., G. BergstroÈm, and I. Groth. 1985. Chemical Outcrossing and a wide dispersal of pollen is mechanically basis for the relationship between Ophrys orchids and their pol- prevented (see above) and by a high degree of odor variation linators. I. Volatile compounds of Ophrys lutea and O. fusca as between individual plants, which is considerably greater than insect mimetic attractants/excitants. Chemica Scripta 25: the variation among ¯owers on single plants. By minimizing 283±311. learned avoidance of the ¯owers, the likelihood that a given Borg-Karlson, A.-K., G. BergstroÈm, and B. Kullenberg. 1987. Chemical basis for the relationship between Ophrys orchids and pollinator would visit several to many different plants within their pollinators. II. Volatile compounds of O. insectifera and a population is therefore increased (Neiland and Wilcock O. speculum as insect mimetic attractants/excitants. Chemica 1995). Density-dependent soft selection or, less likely, group Scripta 27:303±311. selection may be responsible for the evolution of variability Borg-Karlson, A. K., I. Groth, L. AÊ gren, and B. Kullenberg. 1993. Form-speci®c fragrances from Ophrys insectifera L. (Orchida- in scent in individual plants. ceae) attract species of different pollinator genera: evidence of We are currently investigating the pollination biology of sympatric speciation? Chemoecology 4(1):39±45. additional Ophrys species. We do have indications for ¯ower- Buser H.-R., H. Arn, P. Guerin, and S. Rauscher. 1983. Determi- speci®c scent variation in other Ophrys species as well (pers. nation of double bond position in mono-unsaturated acetates by obs.). Nonetheless, based on different ecological and etho- mass spectrometry of dimethyl disul®de adducts. Anal. Chem. 55:818±822. logical constraints including the mating strategies and male Calvo, R. N. 1990. In¯orescence size and fruit distribution among mating behavior of pollinators, we also expect to ®nd repro- individuals in three orchid species. Am. J. Bot. 77:1378±1381. ductive strategies that differ from the one described here for Catling, P. M., and V. R. Catling. 1991. A synopsis of breeding O. sphegodes. Our aim is to reveal the chemical communi- systems and pollination in North American orchids. Lindleyana 6(4):187±210. cation processes involved in the pollinator attraction to obtain Cole, F. R., and D. H. Firmage. 1984. The ¯oral ecology of Pla- information on the evolution of reproductive strategies in tanthera blephariglottis. Am. J. Bot. 71:700±710. orchids of the genus Ophrys. Dafni, A., and P. Bernhardt. 1990. Pollination of terrestrial orchids of southern Australia and the Mediterranean region. Pp. 193± ACKNOWLEDGMENTS 252 in M. K. Hecht, B. Wallace, and R. J. Macintyre, eds. Evo- lutionary biology. Plenum Press, New York. We thank H. Dobson, A. Nilsson, and two anonymous Darwin, C. 1862. On the various contrivances by which British and reviewers for many helpful comments and E. Zellinger and foreign orchids are fertilized by insects. Jon Murray, London. Delforge, P. 1994. Guide des OrchideÂes D'Europe D'Afrique du G. Dutzler for technical assistance. H. Dobson and J. Plant nord et du proche-orient. Delachaux et NestleÂ, Lausanne, Swit- helped to revise the English. We thank the district govern- zerland. ments of Korneuburg and GaÈnserndorf and the Biological Dunkelblum, E, S. H. Tan, and P. J. Silk. 1985. Double-bond lo- Research Institute Illmitz/National Park Neusiedlersee-See- cation in mono-unsaturated fatty acids by dimethyl disul®de. winkel for issuing collecting licenses. This research is sup- Derivatization and mass spectrometry: Application to analysis È of fatty acids in pheromone glands of four Lepidoptera. J. Chem. ported by a grant of the Osterreichische Akademie der Wis- Ecol. 11:265±277. senschaften and the Fonds zur FoÈrderung der wissenschaf- Dutzler, G., and M. Ayasse. 1996. The function of female sex pher- tlichen Forschung Austria (P12275-BIO). Participation of the omones in mate selection of Osmia rufa (Megachilidae) bees. P. Swedish team in this study was supported by grants from the 376 in Proceedings XX Int. Congr. Entomol., Florence, Italy. Swedish Natural Science Research Council and the Knut and Eickwort, G. C., and H. S. Ginsberg. 1980. Foraging and mating behavior in Apoidea. Annu. Rev. Entomol. 25:421±446. Alice Wallenberg Foundation, that of the German team by Eigenbrode, S. D., and K. E. Espelie. 1995. Effects of plant epi- the Deutsche Forschungsgemeinschaft. cuticular lipids on insect herbivores. Annu. Rev. Entomol. 40: 171±194. Evans, H. 1966. The behaviour patterns of solitary wasps. Annu. LITERATURE CITED Rev. Ecol. Syst. 11:123±155. Ackerman, J. D. 1986. Mechanism and evolution of food-deceptive Faegri, K., and L. van der Pijl. 1979. The principles of pollination pollination systems in orchids. Lindleyana 1:108±113. ecology. 3rd ed. Pergamon, Oxford, U.K. Ayasse, M., F. Schiestl, H. F. Paulus, D. Erdmann, and W. Francke. Francke, W. 1986. Convergency and diversity in multicomponent 1997. Chemical communication in the reproductive biology of insect pheromones. Pp. 327±336 in M. Porchet, J.-C. Andries, Ophrys sphegodes. Mitt. Dtsch. Ges. Allg. Angew. Ent. 11: and A. Dhainaut, eds. Advances in invertebrate reproduction. 473±476. Elsevier Science, Amsterdam, The Netherlands. 2006 MANFRED AYASSE ET AL.

Fritz, A.-L., and L. A. Nilsson. 1996. Reproductive success and lution and speciation in Ophrys (Orchidaceae). Israel. J. Bot. 39: gender variation in deceit-pollinated orchids. Pp. 319±338 in D. 43±79. G. Loyd and S. C. H. Barrett, eds. Floral biology: studies on ÐÐÐ. 1994. SignalfaÈlschung als BestaÈubungsstrategie in der med- ¯oral evolution in animal-pollinated plants. Chapman and Hall, iterranen Orchideengattung OphrysÐProbleme der Artbildung New York. und der Artabgrenzung. Eurorchis 92:45±71. Gill, D. E. 1989. Fruiting failure, pollinator inef®ciency, and spe- Peakall, R. 1989. A new technique for monitoring pollen ¯ow in ciation in orchids. Pp. 458±481 in D. Otte and J. A. Endler, eds. orchids. Oecologia 79:361±365. Speciation and its consequences. Sinauer, Sunderland, MA. ÐÐÐ. 1990. Responses of male Zaspilothynnus trilobatus Turner Godfery, M. J. 1925. The fertilisation of Ophrys speculum, O. lutea, wasps to females and the sexually deceptive orchid it pollinates. and O. fusca. J. Bot. (London) 63:33±40. Funct. Ecol. 4:159±167. Heinrich, B. 1975. Bee ¯owers: a hypothesis on ¯ower variety and Peakall, R., and A. J. Beattie. 1996. Ecological and genetic con- blooming times. Evolution 29:325±344. sequences of pollination by sexual deception in the orchid Ca- Hutchings, M. J. 1987. The population biology of the early spider ladenia tentactulata. Evolution 50(6):2207±2220. orchid, Ophrys sphegodes Mill. I. A demographic study from Peakall, R., and S. N. Handel. 1993. Pollinators discriminate among 1975 to 1984. J. Ecol. 75:711±727. ¯oral heights of a sexually deceptive orchid: implications for Knudsen, J. T., and L. Tollsten. 1993. Trends in ¯oral scent chem- selection. Evolution 47(6):1681±1687. istry in pollination syndromes: ¯oral scent composition in moth Priesner, E. 1973. Reaktionen von Riechrezeptoren maÈnnlicher So- pollinated taxa. Bot. J. Linn. Soc. 113:263±284. litaÈrbienen (Hymenoptera, Apoidea) auf Inhaltsstoffe von Kukuk, P. F. 1985. Evidence for an antiaphrodisiac in the sweat OphrysÐBluÈten. Zoon (Suppl. 1):43±54. bee Lasioglossum (Dialictus) zephyrum. Science 227:656±657. Proctor, M., P. Yeo, and A. Lack. 1996. The natural history of Kullenberg, B. 1961. Studies in Ophrys pollination. Zool. Bidr. pollination. Harper Collins, London. Upps. 34-1:1±340. Schiestl, F. P., M. Ayasse, H. F. Paulus, D. Erdmann, and W. ÐÐÐ. 1973. New observations on the pollination of Ophrys Francke. 1997. Variation of ¯oral scent emission and post pol- L.(Orchidaceae). Zoon (Suppl. 1):9±13. lination changes in individual ¯owers of Ophrys sphegodes Kullenberg, B., and G. BergstroÈm. 1976. The pollination of Ophrys subsp. sphegodes (ORCHIDACEAE). J. Chem. Ecol. 23: orchids. Bot. Notiser 129:11±19. 2881±2895. LoÈfstedt, C., W. M. Herrebout, and S. B. J. Menken. 1991. Sex Schiestl, F. P., M. Ayasse, H. F. Paulus, C. LoÈfstedt, B. Hansson, pheromones and their potential role in the evolution of repro- F. Ibarra, and W. Francke. 1999. Orchid pollination by sexual ductive isolation in small ermine moth (Yponomeutidae). Chem- swindle. Nature 399:421±422. oecology 2:20±28. Schiestl, F. P., and M. Ayasse. 2001. Post-pollination emission of McLafferty, F. W., and D. B. Stauffer. 1989. The Wiley NBS reg- a repellent compound in a sexually deceptive orchid: a new istry of mass spectral data. Wiley Interscience, New York. mechanism for maximizing reproductive success? Oecologia. In Moya, S., and J. D. Ackerman. 1993. Variation in the ¯oral fra- press. grance of Epidendrum ciliare (Orchidaceae). Nord. J. Bot. 13: Smith, B. H. 1993. Merging mechanism and adaptation: approach 41±47. to learning and generalization. Pp. 126±157 in D. R. Papaj and Neiland, M. R. M., and C. C. Wilcock. 1995. Maximisation of A. C. Lewis, eds. Insect learning: ecology and evolutionary per- reproductive success by European Orchidaceae under conditions spectives. Chapman and Hall, New York. of infrequent pollination. Protoplasma 187:39±48. Smith, B. H., and M. Ayasse. 1987. Kin-based male mating pref- Nilsson, A. 1980. The pollination ecology of Dactylorhiza sam- erences in two species of halictine bees. Behav. Ecol. Sociobiol. bucina (Orchidaceae). Bot. Notiser. 133:367±385. 20:313±318. ÐÐÐ. 1992. Orchid pollination biology. Trends Ecol. Evol. 7: Sokal, R. R., and F. J. Rohlf. 1995. Biometry. 3rd ed. W. H. Freeman 255±259. and Co., New York. Patt, J. M., M. W. Merchant, D. R. E. Williams, and B. J. D. Meeuse. SPSS. 1997. SPSS base 7.5 for Windows user's guide. SPSS, Inc., 1989. Pollination biology of Platanthera stricta (Orchidaceae) Chicago. in Olympic National Park, Washington. Am. J. Bot. 76: TengoÈ, J. 1979. Odour-released behaviour in Andrena male bees. 1079±1106. Zoon 7:15±48. Paulus, H. F. 1988. Co-Evolution und einseitige Anpassungen in Tollsten, L., and J. BergstroÈm. 1989. Variation and post-pollination BluÈten-BestaÈuber-Systemen. BestaÈuber als Schrittmacher in der changes in ¯oral odours released by Platanthera bifolia (Orchi- BluÈtenevolution. Verh. Dt. Zool. Gesellsch. 81:25±46. daceae). Nord. J. Bot. 9:359±362. Paulus, H. F., and C. Gack. 1986. Neue Befunde zur Pseudoko- Van der Cingel, N. A. 1995. An atlas of orchid pollination: European pulation und BestaÈuberspezi®taÈt in der Gattung OphrysÐUn- orchids. A. A. Balkema, Rotterdam, The Netherlands. van der Pijl, L., and C. H. Dodson. 1966. Orchid ¯owers: their tersuchungen in Kreta, SuÈditalien und Israel. Pp. 48±86 in K. pollination and evolution. Univ. of Miami Press, Coral Gables, Senghas and H. Sundermann, eds. Probleme der Taxonomie, FL. Verbreitung und Vermehrung europaÈischer und mediterraner Or- Westrich, P. 1989. Die Wildbienen Baden-WuÈrttembergs I, II. Eu- chideen II. Sonderheft ``Die Orchidee'', BruÈcke-Verlag Kurt gen Ulmer Verlag, Stuttgart, Germany. Schmersow, Hildesheim, Germany. ÐÐÐ. 1990. Pollinators as prepollinating isolation factors: evo- Corresponding Editor: J. Conner