POLLEN CARRYOVER, GEITONOGAMY, AND THE EVOLUTION OF DECEPTIVE POLLINATION SYSTEMS IN ORCHIDS.
L. A. NILSSON 
(1.) School of Botany and Zoology, University of Natal, Private Bag X01, Scottsville, Pietermaritzburg 3209, South Africa
(2.) Department of Systematic Botany, University of Uppsala, Villavagen 6, S-75236 Uppsala, Sweden
Abstract. Geitonogamy (transfer of pollen among flowers on the same plant) may lead to reduced outcrossing and interfere with sex function. Orchids with pollen packaged into pollinaria would be expected to be particularly vulnerable to the loss of cross-mating opportunities imposed by geitonogamy. We tested the hypothesis that the absence of floral rewards in many orchid species is a means of reducing geitonogamy. Experiments with the deceptive species Orchis mascula and Orchis morio showed that queen bumble bees probe more flowers and stay longer on plants when artificial nectar is added to the flowers. Overall, the data indicated that the evolution of nectar production in deceptive Orchis species would result in moderate to high levels of geitonogamy, as a consequence of the greater number of flowers probed and longer visit duration (60 s) by pollinators. However, the estimated levels of geitonogamy were less than expected, due both to a time delay before freshly withdrawn pollinaria bend into the correct position to strike a stigma and to extensive carryover of pollen. The time elapsed before a freshly withdrawn pollinarium is in the correct position to strike the stigma was found to vary 30-80 s, depending on the orchid species. Since pollinators usually spend [less than]30 s on an inflorescence, we estimate that natural populations of the study species are highly outcrossed. The fraction of the pollen load carried over from flower to flower was found to be 0.67 in O. mascula. Selection should favor longer delays in pollinaria bending and extensive pollen carryover in nectar-producing orchids. This is corroborated by the nectariferous orchid Platanthera chlorantha, which we found to have a pollinaria bending delay of 80 s and a high pollen carryover fraction (0.87). In general, selection for traits that prevent geitonogamy should occur only when pollinators are abundant. Since fruit set of orchids is usually pollinator limited, additional explanations may have to be sought to explain deception. The most pla usible complementary hypothesis is that resources in pollinator-limited orchids are invested in advertising display, rather than nectar production.
Key words: deception; floral traits; geitonogamy; nectar; Orchidaceae; Orchis mascula; Orchis morio; Platanthera chlorantha; pollen carryover; pollen limitation; pollination.
Apparently, plants are able to attract pollinators, while at the same time encouraging them to leave quickly
de Jong et al. (1993)
Most angiosperm flowers produce rewards, e.g., nectar, pollen, oils, resins, or fragrances, which are sought by animal pollinators (Simpson and Neff 1983). Nectar is by far the most common reward in flowers. Larger volumes of nectar have been shown to increase the number of flowers probed by pollinators, as well as the duration of probes, which in turn lead to greater levels of pollen deposition and removal per plant (Thomson and Plowright 1980, Hodges 1981, 1995, Waddington 1981, Zimmerman 1983, Galen and Plowright 1985, Thomson 1986, Pyke et al. 1988, Harder 1990, Mitchell and Waser 1992, Mitchell 1993, Burd 1995).
The seemingly well-established idea that nectar confers fitness benefits is, however, confounded by the existence of thousands of animal-pollinated plant species that produce no rewards in their flowers. The majority of these plants belong to the Orchidaceae. There are an estimated 19 500 orchid species (Dressler 1993), of which more than one-third (8000 species) are deceptive, i.e., do not offer floral rewards to their pollinators (Little 1983, Dafni 1984, Ackerman 1986).
Nonrewarding orchids use a variety of ploys (e.g., sexual deception, brood-site mimicry, Batesian mimicry, and generalized food source deception) to attract pollinators (Little 1983, Dafni 1984, Ackerman 1986, Nilsson 1992, Johnson 1994). Here we are concerned only with the most common type of deception, which involves the exploitation of generalized food-seeking behavior in pollinators. Almost all food source deceptive orchids are pollinated by nectar-seeking insects (Dafni 1984, Ackerman 1986).
Although the pollination biology of many deceptive orchids has been investigated, there has been little progress in determining what selective pressures result in the loss of rewards and why the orchid family should be especially prone to deception (Ackerman 1986, Gill 1989). Shifts between rewarding and deceptive pollination systems have occurred many times in the evolution of the Orchidaceae (Dressler 1981). In the South African genus Disa, for example, there have been at least three evolutionary transitions between nectar-producing and nonrewarding flowers (Johnson et al. 1998).
Most attempts to explain deception have invoked trade-offs between seed production and the energetic costs of nectar (Boyden 1982, Ackerman 1986). According to these arguments, the resources used to produce rewards are costly and better allocated to fruit production and survivorship in some species (cf. South-wick 1984, Ackerman 1986, Ackerman and Montalvo 1990, Pyke 1991). The main problem with this hypothesis is that the lifetime fitness of many orchids is pollination limited, rather than resource limited (Calvo and Horvitz 1990, Calvo 1993).
A totally different approach to the problem was offered by Dressler (1981), who suggested that deception may evolve because of an outcrossing advantage: "if an inflorescence offers food, the bee will work every flower in the inflorescence and probably self-pollinate most of them, unless there are structural devices to prevent this" (Dressler 1981:127). Self-pollination resulting from geitonogamy can reduce fitness by causing inbreeding depression and pollen discounting (loss of pollen that could have been exported to conspecifics) (de Jong et al. 1993, Klinkhamer and de Jong 1993, Barrett and Harder 1996). Thus, selection might favor less or no nectar, if this encourages pollinators to leave the plant rather than visit a long sequence of flowers (cf. Hodges 1995). Despite its plausibility, we are not aware of any tests of the hypothesis that food deception in orchids enhances outcrossing, although Peakall and Beattie (1996) present data showing extensive outcrossing and long distance pollen flow in a sexuall y deceptive orchid species.
The only direct measures of geitonogamy in orchids have involved nectar-producing species. Published estimates of the percentage of pollinaria that are transferred geitonogamously vary from 22% in a Praso-phyllum species (Peakall 1989), to 30% in an Aerangis species (Nillson et al. 1992), and 51% in the worker ant pollinated orchid Microtis panviflora (Peakall and Beattie 1991).
The actual extent of geitonogamy depends on a complex array of factors, including the length of a pollinator visitation sequence, the number of flowers available on a plant, the presence of floral rewards, structural mechanisms such as heterostyly, and the extent of pollen carryover, which is the fraction of the pollen load carried over from one flower to the next in a foraging sequence (Geber 1985, de Jong et al. 1993). Although pollen carryover values have been reported for several plant species with loose pollen (Thomson and Plowright 1980, Galen and Plowright 1985, Thomson 1986, Robertson 1992), we are aware of only one previous attempt to measure pollen carryover in an orchid species (Peakall and Beattie 1991). Orchid pollen is packaged into units known as pollinaria. In some species, the entire pollinarium is deposited intact onto a stigma (i.e., there is no carryover), while in others the pollinarium consists of smaller pollen subunits (known as "massulae"), which progressively break away from pollina ria attached to the pollinator.
Yet another factor that has to be taken into account when estimating geitonogamy in orchids is the phenomenon of pollinarium bending (Darwin 1877). In many orchids, a freshly withdrawn pollinarium must first undergo a bending movement, typically through an angle of 90[degrees], before it is in the correct position to strike the stigma. The time taken to complete this movement varies between 20 s and several hours in some species (Darwin 1877, Dressler 1981). Clearly, geitonogamy will not take place until the bending of the first withdrawn pollinarium has taken place. Darwin viewed pollinarium bending as a mechanism that promotes the production of outcrossed fruits. However, by preventing immediate deposition of self-pollen, pollinarium bending may also promote pollen export, and thus the male function of the flower.
In this study we tested the hypothesis that nonrewarding deception in orchids is a means of reducing geitonogamy. It is known that self-pollination in many orchids, including our study species, leads to reduced seed set (cf. Nilsson 1983, Johnson 1994). Thus, selection should favor traits that reduce the incidence of selfing. To test the hypothesis, we addressed the following questions: (1) Does the number of flowers on a plant probed by a pollinator increase when nectar is added to the flowers of a deceptive species? (2) Does the addition of nectar to flowers cause pollinators to stay on individual plants for longer? (3) How much time must elapse before a freshly withdrawn pollinarium can pollinate another flower on the same plant in deceptive vs. rewarding species? (4) What is the extent of pollen carryover in the study species? (5) How is overall fruit set affected by the addition of nectar to flowers of deceptive species and removal of nectar from flowers of nectar-producing species? And (6) is fruit set in the study species limited by pollinator visits or by other factors?
The study species
Our experiments were carried out on the island of Oland, off the east coast of Sweden. Oland has a rich orchid flora of 30 species. The work for this study was done on Orchis mascula L. and Orchis morio L., which are deceptive species pollinated mainly by queen bumble bees, and Platanthera chlorantha (Cust.) Rchb., a nectar-producing species, which is pollinated by moths. The pollination biology of these orchids has been thoroughly investigated previously (Darwin 1877, Nilsson 1978, 1983, 1984).
Field experiments took place during May-July 1995. The study populations all occur in open "alvar" vegetation and grazed fields within a 5 km radius of the Ecological Research Station of Uppsala University at Skogsby. The populations of O. mascula and O. morio consist of several thousand individuals, while the populations of P. chlorantha consist of 50-100 individual plants.
The study species depend on pollinators for seed set and, like many other orchid species, are partially self-compatible; self pollination typically results in the production of fruits with a reduced percentage of fertile embryos. In O. mascula, for example, the percentage seed set declines from 75.1% when cross-pollinated to 59.8% when self-pollinated (Nilsson 1983).
Pollinator behavior in response to nectar
To determine if pollinators will probe more flowers and spend more time on orchids with nectar, we compared the behavior of queen bumble bees on inflorescences, which had either nectar present or absent. Inflorescences of the deceptive orchids O. mascula and O. morio were used for this experiment. Despite not producing nectar, deceptive Orchis species have floral spurs that act as a convenient repository for artificial nectar.
Since bumble bee visits to natural populations of these species are infrequent and unpredictable, we used a "presentation stick" (cf. Thomson 1988) to place orchid inflorescences near the flight path of foraging or nest-seeking queen bumble bees. This method was found to be effective in inducing the bees to visit the orchids. Bombus lapidarius (L.) was used in all the experiments, as this bee is abundant near the field station and is also one of the primary pollinators of O. mascula and O. morio on Oland (Nilsson 1983, 1984).
The presentation stick consisted of a bamboo rod 2 m in length modified to carry two orchid inflorescences 30 cm apart at either end of a horizontal cross bar. Artificial nectar of 25% sucrose by mass (typical of many bumble bee-pollinated flowers; cf. Percival 1961) was injected by means of a calibrated micro syringe into the floral spurs of all flowers on one inflorescence, while the neighboring inflorescence served as a control. Nectar-enriched and control inflorescences were randomized with respect to left and right positions on the stick. For experiments with O. mascula we added 10 [micro]L of nectar to each flower. In later experiments using O. morio, we added only 2 [micro]L of nectar per flower, as we were interested to determine if small amounts of nectar would also influence pollinator behavior. Inforescences were only used once and were replaced after each visit. Inflorescences consisted of 20 flowers in the case of O. mascula and 12 flowers in the case of O. morio. Where necessary, inflorescence s were trimmed of excess flowers to keep their size constant.
A portable Dictaphone was used to record the number of flowers probed by a bee on each inflorescence, as well as the time spent probing the flowers. We also recorded whether the bee visited both inflorescences on the presentation stick. We placed the two inflorescences at equal distances from each bee, so that we had no control over whether bees would visit the rewarding or control inflorescence.
The behavior of moths on flowers of the rewarding species P. chlorantha was observed in a natural population to determine the number of flowers visited per bout and the average time spent probing each flower. Since the moths that pollinate this species are active only after dusk, a dim flashlight was used to make the observations.
Pollen carryover experiments
To determine the levels of pollen carryover in O. mascula and P. chlorantha, we allowed insects with pollinaria to visit a sequence of virgin emasculated flowers and then recorded the deposition of pollen on the stigmas. Queens of Bombus lapidarius were used for the experiments with O. mascula, and the noctuid moth Autographa gamma (L.) was used for the experiments on P. chlorantha. Both of these insects are important pollinators of the two orchid species on Oland (Nilsson 1978, 1983).
Insects used in the carryover experiments were captured while cruising or foraging on nectar plants (typically Allium schoenoprasum in the case of B. lapidarius and Silene vulgaris in the case of A. gamma) and transferred immediately to a large outdoor flight cage (4 X 4 X 3 m). Any pollinaria already attached to the insects were removed prior to caging. Within one hour of capture, the insect was allowed to visit a single flower containing either two pollinaria (in the case of O. mascula) or one pollinarium (P. chlorantha). Only one pollinarium was used for P. chlorantha because moths typically only remove one pollinarium during a visit (Nilsson 1978). In most cases, the insect readily visited such hand-held flowers. Insects that were reluctant to visit hand-held flowers were released again. Experiments on P. chlorantha were conducted after dusk (2200-2400).
After the insects had visited an initial flower, we waited [greater than or equal to]5 min to allow the freshly affixed pollinaria time to bend into the correct position to effect pollination. The insect was then allowed to visit a sequence of freshly picked flowers that had been bagged with nylon mesh bags prior to anthesis to ensure that their stigmas were not contaminated with pollen. The flowers were also emasculated by removing the pollinaria with forceps before the experiment. We obtained data for seven run sequences in the case of O. mascula and five run sequences for P. chlorantha. A different insect was used for each run. Individual run length was 9-14 flowers (mean, 11.71 flowers, SD = 1.9; n = 7) in the case of O. mascula and 13-31 flowers (mean, 19.6 flowers, SD = 6.2; n = 5) in the case of P. chlorantha.
A permanent marker pen was used to number each flower in the sequence, and flowers were immediately taken back to a laboratory where the number of pollen massulae on the stigma of each flower were counted under a dissecting microscope at 50X. The total number of flowers examined was 82 for O. mascula and 99 for P. chlorantha.
Pollinarium bending rates
According to Darwin (1877), geitonogamy in many orchids may be delayed or even prevented by a "beautiful contrivance": the slow movement of a freshly withdrawn orchid pollinarium into the correct position to strike the stigma. We devised a method for making accurate measurements of the rate of movement in pollinaria. The apparatus consisted of a circular protractor mounted vertically in front of a dissecting microscope that had been placed on its side. A straight piece of wire was attached to the center of the protractor in such a way that it could rotate like the hand of a clock. Freshly withdrawn pollinaria adhering to a toothpick were positioned in the center of the protractor and the bending movement was tracked by the wire, allowing the angle of deflection to be read off from the calibrations on the edge of the protractor. This required two people: one to track the movement of the pollinarium under the microscope, and one to take readings from the protractor at fixed time intervals. Typically, we used 20 pollinaria/species to obtain mean rates of bending.
Addition and removal of nectar in natural orchid populations
To determine if nectar production would increase fruit set in deceptive orchids, we added a single 5 [micro]L dose of 25% sucrose solution to each flower on randomly chosen inflorescences in two populations of Orchis morio. Neighboring inflorescences of a similar size were selected as controls in a paired design. Nectar was added to 961 flowers on 77 plants, and we checked for pollen on stigmas and pollinaria removals in control and experimental plants one week after nectar was added. Fruits were counted several weeks after flowering had finished, although, in one population (Cowmeadow), the mature inflorescences were eaten by cows before fruit set could be determined.
To determine if the loss of nectar production would compromise overall pollination success in a nectar-producing orchid, we used a micro syringe to remove nectar from 438 flowers on 27 plants of P. chlorantha. Nectar was removed on a daily basis at dusk (1900), shortly before moths commenced foraging. This first experiment continued for three weeks from the beginning to end of flowering in the population. Each flower of P. chlorantha secretes nectar for 3.0 d (SD = 0.92; n = 76). The total amount of nectar removed from each flower was 1.85 [micro]L (SD = 0.95; n = 76). The concentration of nectar in this species is 26.9% sugar by mass (SD = 3.5; n = 11). Stigmatic pollen deposition and pollinaria removal were determined for both nectar-depleted and control plants at the end of the experiment.
Because the evolutionary loss of nectar may have a greater effect on pollination success when pollinators are scarce, we performed a second nectar removal experiment under conditions of simulated pollinator limitation. This was done by bagging entire inflorescences of P. chiorantha and exposing them to pollinators for only two days. We then checked stigmatic pollen deposition and pollinaria removal in control vs. nectar-depleted plants.
To determine if fruit set of the deceptive orchid O. mascula was limited by pollen availability, we performed supplemental hand pollination experiments. We hand pollinated as many flowers as possible on randomly selected plants, using pollinaria from [greater than or equal to]5 m away, in order to ensure outcrossing. Later, when flowering was finished, we compared the fruit set of hand-pollinated plants with a set of unmanipulated control plants.
Bumble bees typically probed only 1-2 flowers when visiting control (nonrewarding) inflorescences of O. mascula and O. morio (Fig. 1). This agrees with the observations made by Nilsson (1983) of bumble bee behavior in natural populations of deceptive orchids. Addition of nectar resulted in a significant increase in the mean number of flowers probed per inflorescence, from 1.73 (SD = 0.2; n = 26) to 6.39 flowers (SD = 0.8; n = 23) in O. mascula, and from 1.75 (SD = 0.2; n = 12) to 11.2 flowers (SD = 0.8; n = 8) in O. morio (Fig. 1).
Bees spent a mean of only 7.21 s (SD = 1.8; n = 26) on control inflorescences of O. mascula vs. 57.61 s (SD = 9.6; n = 23) on nectar-enriched inflorescences (Fig. 1). Similarly, addition of nectar significantly increased the mean time spent by bees on inflorescences of O. morio from 5.12 s (SD = 1.6; n = 12) to 53.50 s (SD = 6.5; n = 8) (Fig. 1). The mean time bees spent probing individual flowers of O. mascula increased significantly from 2.49 s (SD = 0.5; n = 26) for control inflorescences to 5.95 s (SD 0.7; n = 23) for inflorescences that had been enriched with 10 [micro]L nectar/flower (Fig. 1). However, there was no significant difference in the time bees spent probing individual flowers of O. mascula on control inflorescences (mean, 2.73 s; SD = 0.3; n = 12) and inflorescences that had been enriched with 2 [micro]L nectar/flower (mean, 2.59 s; SD = 0.6; n = 8) (Fig. 1). Probing time decreased toward the end of foraging bouts on nectar-enriched inflorescences of O. mascula and O. morio (Fig. 2). Mean tra nsit times for bees between flowers of O. mascula were 4.28 s (SD = 3.5; n = 97) on nectar-enriched plants and 4.06 s (SD = 3.5; n = 14) on nectarless plants. Mean transit times for bees between flowers of O. morio were 2.56 s (SD = 1.96; n = 82) for bouts on nectar-enriched plants and 1.77 s (SD = 1.75; n = 6) for bouts on nectarless plants. Interestingly, we observed only two cases where bees flew directly from one inflorescence on the presentation stick to the other inflorescence. Both were bees that had initially probed nectar-enriched flowers.
Despite the production of nectar in the flowers of P. chlorantha, moths probed a mean of only 1.85 flowers/plant (SD = 0.83) during seven bouts observed in a natural population. The total amount of time spent on individual plants was 11.15 s (SD = 7.8; n = 7), the mean time taken to probe each flower was 4.49 s (SD = 3.9; n = 13), and the mean transit time between flowers was 3.53 s (SD = 3.2; n = 5).
The amount of pollen deposited on a sequence of emasculated flowers decreased rapidly in O. mascula and less rapidly in P. chlorantha (Fig. 3). The mean number of flowers reached by pollen was 6.57 flowers (SD = 2.12) for runs involving O. mascula and 13.8 flowers (SD = 2.04) for runs involving P. chlorantha (t = 3.09; P [less than] 0.05). The median pollen grain was transferred a mean distance of 1.42 flowers (SD = 0.78) for runs involving O. mascula and 4.2 flowers (SD = 2.58) for runs involving P. chlorantha (t = 5.38; P [less than] 0.05).
The values for the mean amount of pollen deposited on a sequence of stigmas fitted an exponential decay model better than a linear model for each species; exponential [R.sup.2] = 0.88, linear [R.sup.2] = 0.56 for O. mascula; exponential [R.sup.2] = 0.58, linear [R.sup.2] = 0.48 for P. chlorantha. Carryover proportions obtained from these curves were 0.67 for O. mascula and 0.87 for P. chlorantha (Fig. 3). This means that while 33% of the pollen load is deposited on each sequential flower in O. mascula, only 13% of the pollen load is deposited on each sequential flower in P. chlorantha.
By adding all the massulae deposited on the flowers across a run sequence, we were able to estimate the number of massulae in the initial load of pollinaria (a similar estimate can be obtained directly from the regression). For O. mascula, there are 141 [plus or minus] 47.9 massulae found in both pollinaria. In P. chlorantha, there are 366 [plus or minus] 5.13 massulae in each pollinarium (i.e., 732 massulae/flower). This method assumes that there is no loss of pollen during the sequence, which is reasonable, as the individual pollen massulae are bound together tightly by elastic viscin threads (Darwin 1877). It is only when massulae adhere to a sticky surface, such as the stigma of an orchid flower, that they become forcibly separated from the rest of the pollinarium attached to the insect.
Pollinaria bending times
Pollinaria from different orchid species showed consistent differences in the rate of bending. The time elapsed before the pollinarium has undergone its full angle of movement was 30 s in O. morio, 40 s in O. mascula, and 80 s in P. chlorantha (Fig. 4).
Effect of nectar on fruit set in natural populations
Addition of a single 5 [micro]L dose of nectar to flowers of 0. morio led to significant increases in both pollinaria removal and pollination success after one week in the Stonemeadow population (Table 1). Pollination success was not affected by nectar addition in the Cowmeadow population, which enjoyed a high rate of visitation by bumble bees (Table 1). No significant effect of nectar addition on fruit set was found in the Stonemeadow population (Table 1).
Daily removal of nectar from the flowers of P. chlorantha over the entire flowering season had no significant effect on final levels of pollinaria removal and pollen deposition (Table 2). Removal of nectar resulted in reduced (but not significant) levels of pollinaria removal and pollination in plants that were bagged until all the flowers had reached anthesis and then exposed to pollinators for two days, in a simulation of pollinator limitation (Table 2).
Natural levels of fruit set in the population of O. mascula were very low ([less than]10%). Similarly low fruit set values have been reported for O. morio (Nilsson 1984). Hand pollination increased fruit set to 80% in a population of O. mascula, indicating that fruit set is pollen limited (Table 2).
Our data were consistent with the prediction that nectar results in dramatic increases in both the number of flowers probed by pollinators and the time spent probing flowers. However, as we show, estimated levels of geitonogamy in the study orchids (even those with nectar) are much less than expected, due to a combination of pollen carryover and structural mechanisms, such as pollinarium bending, that prevent immediate self-pollination.
Contrary to expectation, we found that pollen carryover in O. mascula and P. chlorantha is comparable to that in plants with loose pollen (Robertson 1992). We attribute the higher carryover fraction in P. chlorantha to both the higher number of massulae per pollinarium in this species and also the gentler action of moths vs. bumble bees when they probe the flowers.
Estimating geitonogamy in orchids
Although several models of geitonogamy have been developed for plants with loose pollen (Geber 1985, Robertson 1992, de Jong et al. 1993, Barrett and Harder 1996), none of these incorporate a time delay in the onset of geitonogamy. Therefore, we modified the equations of de Jong et al. (1993) so that they can apply to orchids.
Following de Jong et al. (1992, 1993), we denote the fraction of pollen deposited on each flower in a sequence as [k.sub.1]. For application in a discrete model, [k.sub.1] can be estimated as 1 -- [e.sup.-b], where b is the slope in an exponential regression of pollen carryover data (de Jong et al. 1992). Thus [k.sub.1] = 0.28 for O. mascula and 0.12 for P. chlorantha, based on the curves in Fig. 3. The fraction of pollen removed from each flower is denoted as [k.sub.2]. (In orchids, usually [k.sub.2] = 1, unless only one of the two pollinaria are removed per visit, in which case [k.sub.2] = 0.5). Unlike some models for plants with loose pollen (e.g., de Jong et al. 1993), we assume that pollen does not fall off the insect or stick to the corolla. This assumption is reasonable, given that pollinaria are firmly "glued" to the body of the pollinator and individual pollen massulae are held together by elastic viscin threads (Darwin 1877).
The initial load of outcross pollen on the pollinator is A, and the total amount of pollen in each flower is B. We assume that A is equivalent to two pollinaria for bumble bees visiting O. mascula and one pollinarium for moths visiting P. chlorantha. This approximates the number of pollinaria on captured vectors in natural populations (Nilsson 1978, 1983). To make allowance for the pollinaria bending mechanisms of orchids, we used the following parameters: [beta], which is the pollinaria bending time; [[alpha].sub.1], the mean time spent probing each flower; and [[alpha].sub.2], the mean time spent in transit between flowers. Our data showed that the value of [[alpha].sub.1] varies according to the position of a flower in a sequence (Fig. 2), but to make calculations simpler we use the mean value for [[alpha].sub.1] (Fig. 1). The constant [delta] is the number of flowers probed before the first withdrawn pollinarium has undergone its bending movement. We assume that pollen is deposited only once the pollinar ium has undergone its full bending movement. This is an approximation, as the bending movement is gradual, and some deposition may occur before the maximum angle is reached (Fig. 4). The constant [delta] may be calculated as follows:
[delta] = [beta]/[[alpha].sub.1] + [[alpha].sub.2] (1)
By substituting known values for the parameters (Figs. 1, 3, and 4), we can determine that geitonogamy in O. mascula is likely to occur after the pollinator has probed four nectar-enriched flowers, or six flowers not containing nectar. As a consequence of the rapid probing times and slow pollinarium bending rates in P. chlorantha, geitonogamy is likely to occur only after the pollinator has probed nine flowers.
The amount of outcross pollen (C) on the ith flower in a sequence is
C = [Ak.sub.1][(1 - k).sup.i-1] (2)
and the amount of self-pollen (D) on the ith flower in a sequence is
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (3)
We can now substitute known values for O. mascula and plot the expected amounts of outcross and self-pollen on a sequence of flowers (Fig. 5A). Bumble bees probe only two flowers, on average, when visiting non-rewarding inflorescences of O. mascula. Given that the bee arrived with pollen, this will result in two fully outcrossed fruits. On the other hand, bumble bees probe about six flowers on nectar-enriched inflorescences of O. mascula. Each visit will result in four outcrossed fruits and two partially selfed fruits (Fig. 5A). Adding up the cumulative amounts of outcross and self-pollen deposited across a sequence of six flowers, we estimate that the fraction of self-pollination on nectar-enriched plants will be 46.9%. This value may, however, vary considerably if the initial load of pollen A is very different from our estimate, and if the deposition fraction [k.sub.1] varies according to the presence of nectar in flowers. Our assumption that pollen carryover is the same for nectarless and nectar-enriched plants of O. mascula may not be valid if bees deposit more pollen on flowers that contain nectar (cf. Galen and Plowright 1985, Thomson 1986)
Due to the lack of data on pollen carryover in O. morio, we could not derive detailed estimates of geitonogamy in this species. However, we could determine from behavioral data and pollinarium bending rates (Table 1, Fig. 4) that the value of [delta] in this species is 5.8 when nectar is present, meaning that self-pollination will occur from the sixth flower onwards. Since bees visited 11 flowers on nectar-enriched plants, there would be substantial geitonogamy in this species, even if carryover is quite extensive.
Interestingly, our results for P. chlorantha show that nectar production in orchids does not necessarily lead to geitonogamy. The first nine flowers probed by a moth that arrived with pollen should be completely outcrossed (Fig. 5B). This is a consequence of extensive pollen carryover and slow pollinarium bending rates. Since moths probe only a few flowers on each visit to an inflorescence, we estimate that this nectar-producing species is highly outcrossed (Fig. 5B).
By reducing pollen export, geitonogamy can reduce the male fitness of a plant. If a pollinator probes n orchid flowers, the amount of pollen exported (E) is
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4)
In O. mascula, bumble bees will export all the pollen that is removed from the first six flowers when the flowers are nonrewarding. However, when flowers are enriched with nectar, bumble bees will export all the pollen removed from the first four flowers only. The efficiency of pollen export decreases rapidly when visitation sequences are longer than the delay constant [delta].
Since bees typically visit six flowers on nectar-enriched plants of O. mascula, the efficiency of pollen export will be 87% (738 massulae exported vs. 846 removed). On the other hand, since bees visit only two flowers on plants without nectar, the efficiency of pollen export will be 100% (282 massulae exported vs. 282 removed).
The greater amount of pollen removed from plants with nectar will be an advantage when pollinators are scarce; while the greater efficiency of pollen export, when nectar is absent, will be an advantage if pollinators are common enough to remove all the pollen (Klinkhamer and de Jong 1993, Klinkhamer et al. 1994).
Despite producing nectar, the large value of [delta] in P. chlorantha ensures efficient pollen export. Pollen removal in this species will equal pollen export for visitation sequence up to nine flowers in length. Since moths usually visit far fewer than nine flowers on each approach, there should be no pollen discounting, unless pollinaria are misplaced or dislodged during a visit.
Slow pollinaria bending rates, leading to high [delta] values, ensure efficiency of the male function of many orchid flowers. Therefore, pollen export in many orchids can be estimated with reasonable accuracy by counting the number of pollinaria removed from flowers. For those orchids in which small [delta] values lead to geitonogamous pollination, the number of pollinaria removed may not be an accurate estimate of pollen export.
Predictions for other orchids
There is a bewildering array of pollinarium types among orchids (Dressler 1981, 1983). In broad terms, there are two main categories: sectile pollinaria, consisting of numerous individual massulae, and solid pollinaria, which are deposited as a single unit. Pollinaria bending, which we have shown to be an important factor influencing geitonogamy, can occur in both sectile and solid pollinaria. Structural differences between pollinaria may have important implications for evolution of deception.
The orchid species used in our study have sectile pollinaria with a bending mechanism. We refer to this as the "Orchis type" as it is widespread among temperate European genera, such as Orchis, Ophrys, and Platanthera. As a consequence of pollen carryover and the bending mechanism, orchids with this kind of pollinarium do not appear to be especially prone to geitonogamy (Figs. 5 and 6). However, sectile pollinaria lacking a bending mechanism occur in orchids such as Disa, Satyrium, and Habenaria. Orchids with this "Disa type" of pollinaria are more prone to geitonogamy (Fig. 6).
The majority of the world's orchids have solid pollinaria that are deposited onto the stigma as a single unit, thus preventing pollen carryover. Solid pollinaria of the "Oncidium type" have a bending mechanism, which can delay the onset of geitonogamy (Fig. 6), while solid pollinaria of the "Angraecum type" lack a bending mechanism, rendering them most prone to geitonogamy (Fig. 6). If avoidance of geitonogamy has been a major factor in the evolution of deception, then we predict that deception should be most common among orchids with pollinaria of the Angraecum type. A detailed analysis of the correlates of deception in various orchid floras might allow this prediction to be tested.
Nectar and fruit set
Deceptive orchid species usually have low levels of fruit set, while rewarding orchid species in the same habitats often enjoy high levels of fruit set (cf. Dafni and Ivri 1979, Gill 1989, Johnson and Bond 1997). This pattern, together with the observations that bumble bees visited more flowers on nectar-enriched plants (Fig. 1) and that pollination success was boosted a week after nectar addition in a population of O. morio (Table 1), led us to expect that artificially nectar-enriched plants would enjoy higher levels of fruit set. Thus it was surprising that nectar addition did not lead to significant increases in fruit set in a population of O. morio (Table 1). However, we suspect that the results may have been different if nectar had been added on a daily basis, rather than in a single dose that may have quickly evaporated or decomposed. Nevertheless, some orchid flowers secrete nectar only at the beginning of anthesis (cf. Ackerman et al. 1994), so the addition of a single dose of nectar may not be unrealistic.
Nectar removal from the flowers of P. chlorantha was carried out on a daily basis, yet even this treatment did not significantly affect pollination success and pollinaria removal, even when pollinator availability was artificially restricted (Table 2). We observed that moths seldom visit many flowers in a sequence, even when nectar is present, so it would appear that nectar has little influence on the behavior of moths once they have alighted.
Other studies also indicate that the presence of nectar in flowers does not automatically lead to higher levels of fruit set. Burd (1995) showed that artificially enhanced nectar rewards in Lobelia deckenii would not alleviate pollen limitation, and Lopez-Portillo et al. (1993) found no difference in fruit set between nectarless and nectar-producing morphs of Prosopis glandulosa. Ackerman (1981) found that nectar addition did not affect fruit set of the deceptive orchid Calypso bulbosa, although the artificial nectar used in his experiment was very dilute (10% sugar by mass) for bumble bees. Unfortunately, addition of nectar with a more realistic concentration (30%) caused the flowers of this species to wilt (Ackerman 1981). Cutting the tip off spurs to prevent nectar production in the hummingbirdpollinated orchid Comparettia falcata led to reduced fruit set only in some populations (Ackerman et al. 1994).
The very low levels of pollination and fruit set in the deceptive Orchis species in this study makes us doubt that deception has evolved in these species primarily as a means to promote outcrossing. According to theory, traits that promote outcrossing should be under strong selection only when pollinators are abundant and fruit set is resource limited (de Jong et al. 1993).
A general survey of the literature shows that fruit set in deceptive orchids is nearly always low (Gill 1989). This is apparently due to chronic pollen limitation, as supplemental hand pollination usually results in high levels of fruit set (cf. Table 2). The challenge therefore is to think of plausible reasons why nectarproducing mutants have not invaded populations of deceptive orchids (Gill 1989), Presumably there must be circumstances in which the evolution of nectar production does not lead to increased lifetime seed production.
Under conditions of pollen limitation, we would expect selection to favor traits that increase the attractiveness of the plant to pollinators and/or encourage pollinators to visit several flowers in succession, even if the quality of fruits is compromised through geitonogamy. Many orchids are small herbs that struggle to compete for the attraction of pollinators, therefore resources may be better allocated to a larger floral display for long-distance attraction of pollinators than to floral rewards that influence pollinator behavior after alighting.
In natural populations of deceptive orchids, many plants do not set any fruit at all. In populations of O. mascula, for example, 50% of individual plants do not set fruit in any one year (Nilsson 1983). The main difficulty facing these orchids seems to be the initial attraction of pollinators. We turned to previously unpublished data on fruit set in O. morio to examine the relationship between floral display and reproductive success. There is a good correlation between the number of flowers and fruits on a plant (Fig. 7A), but this could simply mean that pollinators visit more flowers on large plants. To test the hypothesis that plants with many flowers are more successful at attracting pollinators, we regressed the probability that a plant will set at least one fruit against flower number. The results show clearly that small plants are more likely to be completely neglected by pollinators (Fig. 7B). From this we conclude that selection will strongly favor plants with a superior floral display.
Unfortunately, there are few reliable estimates of the cost of nectar (Pleasants and Chaplin 1983, Pyke 1991). Estimates of the cost of nectar production range from just 3.3% of flower costs in the short-lived flowers of Pontederia cordata (Harder and Barrett 1992), to 30% of flower costs in the long-lived flowers of Asclepias quadrifolia and A. syriaca (Pleasants and Chaplin 1983, Southwick 1984). Given that orchid flowers are long lived, making continuous nectar production expensive, these figures suggest that deception may allow plants to increase the number of flowers per inflorescence by up to 30%, which may substantially improve a plant's chances of being noticed by a pollinator (Fig. 7).
On the other hand, there is evidence that even very modest amounts of nectar can have marked effects on the number of flowers probed by pollinators (Ackerman et al. 1994, Burd 1995, Hodges 1995). For example, addition of only 2 [micro]L nectar/flower greatly increased probing by bumble bees in O. morio (Fig. 1). In terms of cost arguments, it is difficult to explain why Orchis species do not produce even a very small amount of nectar.
This study was made possible by a Foundation for Research and Development Postdoctoral Scholarship to S. D. Johnson. We thank Kathy Johnson for much help in the field and the staff of the Ecological Field Station of Uppsala University on Oland for their assistance in many ways. We are also grateful to William Bond, Spencer Barrett, Tom de Jong, James Thomson, and two anonymous reviewers for helpful comments on the manuscript. Paul Husler (University of Cape Town) is thanked for his help in deriving the equation to estimate pollen export.
Ackerman, J. D. 1981. Pollination biology of Calypso bulbosa var. occidentalis (Orchidaceae): a food-deception system. Madrona 28:101-110.
-----. 1986. Mechanisms and evolution of food-deceptive pollination systems in orchids. Lindleyana 1:108-113.
Ackerman, J. D., and A. M. Montalvo. 1990. Short- and long-term limitations to fruit production in a tropical orchid. Ecology 71:263-272.
Ackerman, J. D., J. A. Rodriguez-Robles, and E. J. Meldndez. 1994. A meager nectar offering by an epiphytic orchid is better than nothing. Biotropica 26:44-49.
Barrett, S. C. H., and L. D. Harder. 1996. Ecology and evolution of plant mating. Trends in Ecology and Evolution 11:73-79.
Boyden, T. C. 1982. The pollination biology of Calypso bulbosa var. americana (Orchidaceae): initial deception of bumble bee visitors. Oecologia 55:178-184.
Burd, M. 1995. Pollinator behavioural responses to reward size in Lobelia deckenii: no escape from pollen limitation of seed set. Journal of Ecology 83:865-872.
Calvo, R. N. 1993. Evolutionary demography of orchids: intensity and frequency of pollination and the cost of fruiting. Ecology 74:1033-1042.
Calvo, R. N., and C. C. Horvitz. 1990. Pollinator limitation, cost of reproduction, and fitness in plants, a transition matrix demographic approach. American Naturalist 136:499-516.
Dafni, A. 1984. Mimicry and deception in pollination. Annual Review of Ecology and Systematics 15:259-278.
Dafni, A., and Y. Ivri. 1979. Pollination ecology of, and hybridization between, Orchis coriophora L. and O. collina Sol. ex Russ. (Orchidaceae) in Israel. New Phytologist 83:181-187.
Darwin, C. H. 1877. On the various contrivances by which British and foreign orchids are fertilized by insects. John Murray, London, UK.
de Jong, T. J., P. G. L. Klinkhamer, and M. J. van Staalduinen. 1992. The consequences of pollination biology for selection of mass or extended blooming. Functional Ecology 6:606-615.
de Jong, T. J., N. M. Waser, and P. G. L. Klinkhamer. 1993. Geitonogamy: the neglected side of selfing. Trends in Ecology and Evolution 8:321-325.
Dressler, R. 1981. Orchids--natural history and classification. Harvard University Press, Cambridge, Massachusetts, USA.
-----. 1993. Phylogeny and classification of the orchid family. Dioscorides Press, Portland, Oregon, USA.
Galen, C., and R. C. Plowright. 1985. The effects of nectar level and flower development on pollen carryover in inflorescences of fireweed (Epilobium angustifolium) Onagraceae. Canadian Journal of Botany 63:488-491.
Geber, M. A. 1985. The relationship of plant size to self-pollination in Mertensia ciliata. Ecology 66:762-772.
Gill, D. E. 1989. Fruiting failure, pollinator inefficiency and speciation in orchids. Pages 456-481 in D. Otte and J. A. Endler, editors. Speciation and its consequences. Sinauer, Sunderland, Massachusetts, USA.
Harder, L. D. 1990. Behavioral responses by bumble bees to variation in pollen availability. Oecologia 85:41-47.
Harder, L. D., and S. C. H. Barrett. 1992. The energy cost of bee pollination for Pontederia cordata (Pontederiaceae). Functional Ecology 6:1-7.
Hodges, C. M. 1981. Optimal foraging in bumble bees: hunting by expectation. Animal Behaviour 29:1166-1171.
Hodges, S. A. 1995. The influence of nectar production on hawkmoth behavior, self-pollination, and seed production in Mirabilis multiflora (Nyctaginaceae). American Journal of Botany 82:197-204.
Johnson, S. D. 1994. Evidence for Batesian mimicry in a butterfly pollinated orchid. Biological Journal of the Linnean Society 53:91-104.
Johnson, S. D., and W. J. Bond. 1997. Evidence for widespread pollen limitation of fruiting success in Cape wildflowers. Oecologia 109:530-534.
Johnson, S. D., H. P. Linder, and K. E. Steiner. 1998. Phylogeny and adaptive radiation of pollination systems in Disa (Orchidaceae). American Journal of Botany 85:402-411.
Klinkhamer, P. G. L., and T. de Jong. 1993. Attractiveness to pollinators: a plant's dilemma. Oikos 66:180-184.
Klinkhamer, P. G. L., T. de Jong, and A. J. Metz. 1994. Why plants can be too attractive--a discussion of measures to estimate male fitness. Journal of Ecology 82:191-194.
Little, R. J. 1983. A review of floral food deception mimicries with comments on floral mutualism. Pages 294-309 in C. E. Jones and R. J. Little, editors. Handbook of experimental pollination biology. Van Nostrand Reinhold, New York, New York, USA.
Lopez-Portillo, J., L. E. Eguiarte, and C. Montana. 1993. Nectarless honey mesquites. Functional Ecology 7:452-461.
Mitchell, R. J. 1993. Adaptive significance of Ipomopsis aggregata nectar production: observation and experiment in the field. Evolution 47:25-35.
Mitchell, R. J., and N. M. Waser. 1992. Adaptive significance of Ipomopsis aggregata nectar production: pollination success of single flowers. Ecology 73:633-638.
Nilsson, L. A. 1978. Pollination ecology and adaptation in Platanthera chlorantha (Orchidaceae). Botaniska Notiser 131:35-51.
-----. 1983. Anthecology of Orchis mascula (Orchidaceae). Nordic Journal of Botany 3:157-179.
-----. 1984. Anthecology of Orchis morio (Orchidaceae) at its outpost in the north. Nova Acta Regiae Societatis Scientiarum Upsaliensis. Series V, volume C 3:167-179.
-----. 1992. Orchid pollination biology. Trends in Ecology and Evolution 7:255-259.
Nilsson, L. A., E. Rabakonandrianina, and B. Pettersson. 1992. Exact tracking of pollen transfer and mating in plants. Nature 360:666-667.
Peakall, R. 1989. A new technique for monitoring pollen flow in orchids. Oecologia 79:361-365.
Peakall, R., and A. J. Beattie. 1991. The genetic consequences of worker ant pollination in a self-compatible clonal orchid. Evolution 45:1837-1848.
Peakall, R., and A. J. Beattie. 1996. Ecological and genetic consequences of pollination by sexual deception in the orchid Caladenia tentactulata. Evolution 50:2207-2220.
Percival, M. 1961. Floral Biology. Pergamon Press, Oxford, UK.
Pleasants, J. M., and S. J. Chaplin. 1983. Nectar production rates of Asclepias quadrifolia: causes and consequences of individual variation. Oecologia 59:232-238.
Pyke, G. H. 1991. What does it cost a plant to produce floral nectar? Nature 350:58-59.
Pyke, G. H., L. P. Day, and K. A. Wale. 1988. Pollination ecology of Christmas Bells (Blandfordia nobilis Sm.): effects of adding artificial nectar on pollen removal and seed set. Australian Journal of Ecology 13:279-284.
Robertson, A. W. 1992. The relationship between floral display size and geitonogamy in Myosotis colensoi (Kirk) MacBride (Boraginaceae). Biological Journal of the Linnean Society 46:333-349.
Simpson, B. B. and J. L. Neff. 1983. Evolution and diversity of floral rewards. Pages 142-159 in C. E. Jones and R. J. Little, editors. Handbook of experimental pollination biology. Van Nostrand Reinhold, New York.
Southwick, E. E. 1984. Photosynthate allocation to floral nectar: a neglected energy investment. Ecology 65:1775-1779.
Thomson, J. D. 1986. Pollen transport and deposition by bumble bees in Erythronium: influences of floral nectar and bee grooming. Journal of Ecology 74:329-341.
-----. 1988. Effects of variation in inflorescence size and floral rewards on the visitation rates of traplining pollinators of Aralia hispida. Evolutionary Ecology 2:65-76.
Thomson, J. D., and R. C. Plowright. 1980. Pollen carryover, nectar rewards, and pollinator behavior with special reference to Diervilla lonicera. Oecologia 46:68-74.
Waddington, K. D. 1981. Factors influencing pollen flow in bumble bee-pollinated Delphinium virescens. Oikos 37:153-159.
Zar, J. H. 1984. Biostatistical analysis. Prentice Hall, Englewood Cliffs, New Jersey, USA.
Zimmerman, M. 1983. Plant reproduction and optimal foraging: experimental nectar manipulations in Delphinium nelsonii. Oikos 41:57-63.
Manuscript received 9 July 1997; revised 30 June 1998; accepted 25 September 1998; final version received 18 November 1998.
(3.) E-mail: Johnsonsd@botany.unp.ac.za
Effects of adding artificial nectar to flowers or supplemental hand pollination on fruit set of orchids. No. flowers/plant Control Treatment t P Addition of nectar to O. morio flowers at Stonemeadow site [+] 13.5 [plus or minus] 2.4 12.9 [plus or minus] 2.4 1.18 0.23 (62) (62) Hand pollination of O. mascula flowers [++] 8.8 [plus or minus] 2.3 10.3 [plus or minus] 4.6 1.27 0.21 (25) (13) No. flowers/plant Fruit set (%) Control Control Addition of nectar to O. morio flowers at Stonemeadow site [+] 13.5 [plus or minus] 2.4 10.0 [plus or minus] 11.4 (62) (62) Hand pollination of O. mascula flowers [++] 8.8 [plus or minus] 2.3 8.1 [plus or minus] 19.1 (25) (25) No. flowers/plant Control Treatment Z Addition of nectar to O. morio flowers at Stonemeadow site [+] 13.5 [plus or minus] 2.4 14.9 [plus or minus] 21.6 0.85 (62) (62) Hand pollination of O. mascula flowers [++] 8.8 [plus or minus] 2.3 80.4 [plus or minus] 15.8 5.06 (25) (13) No. flowers/plant Control P Addition of nectar to O. morio flowers at Stonemeadow site [+] 13.5 [plus or minus] 2.4 0.39 (62) Hand pollination of O. mascula flowers [++] 8.8 [plus or minus] 2.3 [less than]0.001 (25) Note: Values are expressed as means [plus or minus] 1 SD, with n (no. plants) shown in parentheses. (+.)Treatment consisted of the addition of 5 [micro]L of artificial nectar (25% sugar by mass) to all the flowers. The Z statistic refers to the normal approximation to the Wilcoxon paired sample test (Zar 1984). (++.)The Z statistic refers to the normal approximation to the Mann-Whitney U test (Zar 1984).
|Printer friendly Cite/link Email Feedback|
|Author:||JOHNSON, S. D.; NILSSON, L. A.|
|Date:||Dec 1, 1999|
|Previous Article:||BUMBLE BEE SELECTION OF MIMULUS GUTTATUS FLOWERS: THE EFFECTS OF POLLEN QUALITY AND REWARD DEPLETION.|
|Next Article:||ANT-MEDIATED SEED DISPERSAL ALTERS PATTERN OF RELATEDNESS IN A POPULATION OF TRILLIUM GRANDIFLORUM.|