|Home | About | Journals | Submit | Contact Us | Français|
Pollen fates strongly influence mating success in plants but are difficult to quantify. By promoting foraging constancy in pollinators, floral rewards such as nectar may enhance the overall efficiency of pollen transfer. However, this can also lead to high levels of geitonogamy. Pollen fates were studied in Acrolophia cochlearis, a member of a terrestrial epidendroid orchid genus that includes both rewarding and deceptive species.
Pollinator observations were conducted. Pollen transfer efficiency (PTE), the proportion of removed pollinia deposited on stigmas, was measured in a large population at regular intervals throughout the 5-month flowering season. The level of cross-pollination in two populations was estimated from the percentage of seeds with embryos in naturally pollinated fruits.
Acrolophia cochlearis (and a congener A. micrantha) produce minute but concentrated nectar rewards. Observations showed that A. cochlearis is pollinated exclusively by a solitary bee species, Colletes claripes. Although both sexes visited flowers, only males carried pollinaria. Overall levels of pollination and PTE of the rewarding A. cochlearis were much higher than in a deceptive congener, A. capensis. Seeds resulting from self-fertilization had a significantly lower probability of containing viable embryos than did those from cross-fertilization. This dichotomy in fruit quality was used to estimate that cross-pollination occurred in approx. 66 % of A. cochlearis flowers in a large dense population and approx. 10 % in a small sparse population. Traits of A. cochlearis that limit geitonogamy include pollinarium reconfiguration that exceeds the visit time of pollinators and rapid flower senescence following visitation.
Presence of a nectar reward in Acrolophia cochlearis results in high levels of PTE. It is estimated that approx. 33–90 % of fruits in natural populations arise from self-pollination in this species.
Nectar rewards encourage foraging constancy by pollinators (Goulson, 1999) and, as a result, should be expected to increase pollen transfer efficiency (PTE). PTE can be considered a population-level measure of the proportion of pollen removed from anthers that is subsequently deposited on conspecific stigmas (Johnson et al., 2005; Harder and Johnson, 2008). In general, adaptations that increase PTE should enhance male fitness of individuals.
Nectar rewards can also increase the number of flowers that are visited by individual pollinators on a plant (cf. Johnson et al., 2004). This can result in high levels of geitonogamy (Charlesworth and Charlesworth, 1987b; Johnson et al., 2004; Jersáková and Johnson, 2006), which in turn may be detrimental to fitness because of the production of inferior, inbred offspring (Darwin, 1878; Charlesworth and Charlesworth, 1987a) and squandering of pollen grains and ovules that could otherwise be used for cross-pollination, processes known as pollen and ovule discounting, respectively (Barrett, 2002).
Patterns of pollen dispersal and geitonogamy are notoriously difficult to quantify. For plants that do not have aggregated pollen, tracking pollen movement is limited to the use of fluorescent powders that serve as pollen analogues (Snow et al., 1996) as well as molecular markers that estimate parentage of offspring (e.g. Ritland, 1986; Eckert and Barrett, 1994; Galloway et al., 2003; Herlihy and Eckert, 2004; Kruszewski and Galloway, 2006; Kropf and Renner, 2008). In orchids and asclepiads, packaging of pollen into pollinia allows for more direct tracking of pollen movement by labelling with unique microtags (Nilsson et al., 1992), histochemical stains (Folsom, 1987; Peakall, 1989; Nilsson et al., 1992; Salguero-Faría and Ackerman, 1999; Johnson et al., 2005), coloured powders (Kropf and Renner, 2008) or radioisotopes (Pleasants, 1991).
This study focuses on the reproductive biology and patterns of pollen dispersal in Acrolophia, a poorly known and enigmatic genus of seven terrestrial species representing the only ‘epidendroid’ clade centred in the Cape (Linder and Kurzweil, 1999). Acrolophia is generally thought to be closely aligned to Eulophia (Linder and Kurzweil, 1999; although see Dressler 1993), a large African genus of deceptive terrestrial orchids. To date, nothing has been documented regarding the reproductive biology in the genus, although Russell (2005) observed high rates of seed set in A. cochlearis and consequently suggested that this species might be autogamous.
During the course of this study, it became apparent that A. cochlearis and A. micrantha produce nectar. This is an unusual condition among terrestrial epidendroid orchids in South Africa –nectar production has not been found in any of the many species of Eulophia examined (Peter and Johnson, 2006a, 2008; Peter, 2008). This provided a unique opportunity to study implications of nectar production for pollinia fates.
Although this study focuses on the pollination biology of Acrolophia cochlearis, it also considers aspects of the reproductive biology of A. micrantha and A. capensis. The six objectives were: (1) to determine the identity of pollinators and their interactions with flowers; (2) to determine the volume and concentration of nectar rewards; and (3) to determine the natural visitation rates to flowers and pollen transfer efficiencies of these rewarding and deceptive species. Given the production of nectar in A. cochlearis and numerous flowers on a plant, which both increase potential for geitonogamy, characteristics such as pollinarium reconfiguration that might function to reduce levels of geitonogamy were investigated (4). Finally, levels of inbreeding depression were quantified in a breeding system experiment (5) and the resulting difference in seed quality used to estimate levels of self- and cross-pollination in naturally pollinated flowers (6).
This study was undertaken mainly in a large population (approx. 1500 individuals) of Acrolophia cochlearis (Lindl.) growing along the edges of Mountain Drive, a dirt road along the crest of the Rietberg, south of Grahamstown, Eastern Cape, South Africa. A small sparse population of A. cochlearis growing on a sandy road cutting near the town of Kenton-on-Sea, 35 km south, south-east of Grahamstown was also examined. Additional observations were made on the ridge about 1·5 km to the south of Mountain Drive where the congeneric A. capensis is common and at the Coega rescue nursery near Port Elizabeth where plants of A. cochlearis and A. micrantha had been transplanted a short distance from a site being developed.
This study focuses on A. cochlearis, a relatively common orchid occurring from the Cape Peninsula in the west to the vicinity of East London in the east, with further sparse occurrences up the east coast of South Africa to about Richards Bay. This species produces extensive, branching inflorescences with numerous small non-resupinate flowers (Fig. 1A, B). Flowers are mostly drab and inconspicuous, with a white to pale cream labellum and a short sac-like spur. When numerous flowers are enclosed in a bottle to concentrate their scent, they have an obvious unpleasant sweet scent, although it is difficult to make out the scent of individual flowers or a few flowers on an inflorescence. For all measurements described below, only a single flower was sampled per plant.
Acrolophia micrantha is similar in many respects to A. cochlearis, although the flowers are resupinate and the labellum is larger and less ‘clam-shell-like’ (Fig. 2A). This species also has a short, broad sac-like spur.
In contrast, branched inflorescences of A. capensis (Fig. 2B) are more compact, and most of flowers open within a short period, remaining open for about 2 weeks. Sepals and petals are less spreading than in the case of the previous two species and form, with the base of the labellum, a tube surrounding the column. The central lobe of the labellum is large and showy, being bright white. Flowers of plants from around Grahamstown have a sweet honey-like scent.
Observations of pollinators were conducted at all three study sites on all three study species. However, only visits to A. cochlearis were observed at various sub-populations along Mountain Drive. Unlike in the case of many deceptive orchids where pollinators bearing pollinaria are typically captured on rewarding plants in the surrounding habitat, pollinators were only observed visiting A. cochlearis flower or patrolling clumps of these plants. A total of approx. 38 h was spent observing pollinators on A. cochlearis in 2003 and 2006/2007.
Insects were collected, killed using an ethyl acetate killing jar and mounted for identification. These are housed in the collection of the first author with vouchers housed in the entomology collection in the Albany Museum, Grahamstown (AMGS) and in the collection of Dr Michael Kuhlmann, Institute of Landscape-Ecology, University of Münster, Münster, Germany. Vouchers of plant species are housed in the Schonland Herbarium, Grahamstown (GRA).
Attempts were made to observe pollinators either visiting flowers or patrolling in the vicinity of A. micrantha and A. capensis plants, but these attempts were unsuccessful. Approximately 10 h were spent observing each of the other two species.
Nectar rewards for the three species were investigated. An examination of flowers from at least 15 A. capensis plants failed to detect nectar. The other two species contained minute quantities of nectar in the sac-like spurs. Flowers were therefore carefully broken open to reveal the nectar droplet. This was sucked up into a 4-μL micropipette. The length of the nectar column inside the pipette was measured using electronic calipers, and its volume then determined from this length relative to the length of the calibrated portion of the micropipette, which corresponds to a known volume. Because of the minute quantities of nectar it was not possible to measure nectar concentrations directly. Each sample was therefore diluted with a specific volume of water (also measured accurately with callipers in the same micropipette as the sample). The diluted nectar solution was then placed on the stage of an Atago 50 % sucrose refractometer and the concentration recorded. Knowing the dilution factor, it was possible to calculate the original nectar concentration.
Nectar was collected onto filter paper following refractometer measurements. These samples were used to determine the ratio of fructose : glucose : sucrose using high performance liquid chromatography as described by van Wyk et al. (1993). Because of minute volumes in each flower, samples from a number of individual flowers were pooled on the filter paper for later analysis.
Weekly surveys were conducted in a large population of A. cochlearis growing along the verge of a section of Mountain Drive. One flower was sampled randomly from each of approx. 160 plants per week. In addition, the numbers of plants not yet in flower and those that had completed flowering were counted.
Flowers collected were scored for pollinarium removal, pollinia deposition and failed visits (flowers with their anther cap disturbed but pollinaria not removed). From these data, as well as PTE, percentages were calculated of (a) plants in flower; (b) flowers with their pollinaria removed; (c) flowers pollinated; (d) flowers showing any sign of legitimate visitation (removal, deposition or both); and (e) failed visits. PTE is the percentage of removed pollinia (pollinaria multiplied by two) that are deposited on stigmas (Johnson et al., 2004). In addition, a survey of PTE and other measurements of pollination success, as listed above, were made on deceptive A. capensis.
Due to the small size of pollinaria, bending rates of pollinaria in this species were measured under a dissecting microscope. Pollinaria were removed from flowers on the head of a pin and positioned above a protractor. Angles were then measured and plotted against the time after removal. Reconfiguration was assumed to be complete once the angle stopped changing. Visual inspections were used to determine duration of the pollinator visits to inflorescences.
While examining natural visitation rates, it was noticed that flowers that had pollinaria removed started closing. To document this, pollinaria from flowers were removed and the span of the lateral sepals as well as width of the labellum measured before and 48 h after treatment. Control flowers on the same inflorescence were marked and measured at the same times. Percentage change data were arcsine-square root (angular) transformed to meet assumptions of normality required for ANOVA.
Bagged flowers were cross-pollinated with pollinia from a plant >5 m away, self-pollinated, or left unmanipulated to test for auto-pollination. Hybrid crosses between A. cochlearis plants and a single individual of A. micrantha were also performed. Approximately 16 weeks after pollination we determined the proportion of flowers in the treatment groups that set fruit and percentage of seeds with viable embryos in each fruit. Percentage data were arcsine-square root transformed.
We scored a sample of approx. 150 seeds in naturally produced fruit for the percentage of seeds with embryos. Using results of the breeding system as a reference, the proportion of seeds with embryos was used to determine whether a fruit resulted from self- or cross-pollination.
The observations indicate that A. cochlearis is pollinated exclusively by male Colletes claripes bees (Colletidae; Fig. 1C). These bees accumulate large clumps of tiny pollinaria on their clypei (lower margin of the face), forming large yellow masses (Fig. 1D) clearly visible even on insects in flight. The average number of pollinaria per male bee was 14 with a range of 0–25 (n = 8). Besides the nine captured male bees, a further eight bees seen to be carrying pollinaria masses were not captured in the 2003 and 2006/2007 seasons. These were presumed to be male, given their colour.
Two female C. claripes bees were also observed visiting flowers, but neither carried pollinia. Females were relatively rarely encountered, whereas males apparently stay in the vicinity of a patch of plants, possibly for many days. However, as no insects were tagged, it is not possible to tell whether insects encountered at certain sites on consecutive days were the same individuals.
Both male and female bees when alighting on a flower assume an upside-down position and probe the short, sac-like spur for nectar (Fig. 1C).
A number of flies and occasional vespid wasps and honey bees were seen probing for nectar, but none of these insects removed pollinaria.
Acrolophia cochlearis has minute quantities of nectar with an average volume of only 0·037 µL (s.e. = 0·003; n = 36) per flower. The nectar is concentrated with an average sugar concentration of 90 % (s.e. = 0·467; n = 32). The average ratio of fructose : glucose : sucrose for a number of pooled samples from different plants in two separate analyses was 14 : 16 : 70. In A. micrantha, flowers contain 0·045 µL of nectar at a sugar concentration of 71 % (n = 2). The sugar composition of A. micrantha nectar was not analysed. None of the flowers from 15 individuals of A. capensis in the Rietberg population near Grahamstown produced nectar.
The flowering period in the A. cochlearis population extends over a 5-month period. It was possible to observe rates of flower visitation in the study population near Grahamstown at weekly intervals throughout the flowering period. By the fourth week of flowering, approx. 90 % of the plants were in flower, and flowering continued at this level for another 10 weeks before declining (Fig. 3A).
Overall visitation (flowers showing signs of pollen removal or deposition) increased steadily over the course of the flowering period (Fig. 3A), as did rates of pollinarium removal and pollinia deposition (Fig. 3B). PTE, on the other hand, fluctuated markedly throughout the flowering period (Fig. 3B). PTE in this species reaches 60 % in 2 weeks during the middle of the flowering period. High rates of PTE at the start of the flowering period may be a result of relatively small sample sizes early in the season and few flowering plants in the population. Pollination failure rates also fluctuated throughout the season but were highest in the first few weeks (Fig. 3A).
Freshly removed pollinaria remain erect, perpendicular to the surface of the clypeus and in a position where the pollinia are unlikely to contact the relatively large and exposed stigma. The pollinarium bends forward until it is apressed against the clypeus, a position where it is possible for the pollinia to be deposited on the stigma of a subsequent flower. This bending is comparable to the ‘depression’ of pollinaria Darwin described in a number of European orchids (Darwin, 1867) and takes 88 s on average (s.e. = 5·25; n = 15). The average visit time to inflorescences by male Colletes bees was 28 s (n = 3 visits by three bees to three inflorescences).
Pollinaria of A. micrantha undergo a similar reconfiguration, although reconfiguration time was not recorded in this species. Pollinaria of A. capensis on the other hand show no sign of reconfiguration, although anther cap retention similar to that described in Eulophia foliosa (Peter and Johnson, 2006a) has been suggested for this species in the west of its range (Russell, 2005).
After removal of pollinia, flowers of A. cochlearis undergo a rapid change. Petals and sepals close around the column, and the labellum folds around its long axis and becomes stiff with less articulation at the mentum (Table 2). The labellum also rapidly changes from white or pale cream to a darker yellow-cream and ultimately to the brown of sepals and petals. Because of this rapid change, in many plants there are only one or two fresh flowers open per inflorescence branch at any one time.
Control flowers typically remained unchanged for longer than 48 h, but because random pairs of control and experimental flowers were chosen, some of them had been on the inflorescences longer than others and so after about 60 h some control flowers were beginning to close as a result of age.
There was no difference in rates of fruit set between cross- and self-pollinated flowers (Table 3). However, the weight of capsules from self-pollination was significantly lower than that from cross-pollination and hybrid cross-pollination, as was percentage of fertile seeds in a fruit (Table 3). Bagged, unmanipulated flowers failed to set fruit, ruling out auto-pollination in this species.
The different signals of fertile seeds produced by experimentally self- and cross-pollinated flowers (Table 3 and Fig. 4) were used to estimate the frequency of cross- and self-pollination in naturally pollinated flowers. Based on this experiment, fruits with ≤60 % of their seeds containing embryos were considered to result from self-pollination, and those with ≥61 % of their seeds containing embryos were considered to result from cross-pollination (Fig. 4). This analysis indicated that in a large and dense population, a high proportion (66 %) of naturally produced fruit is likely to be a result of cross-pollination (Fig. 4). In contrast, 94 % of naturally produced fruit in a small sparse population is likely to be the product of self-pollination (Fig. 4).
Acrolophia contains both rewarding and deceptive species, in contrast to Eulophia, a large putatively related genus that appears to contain only deceptive species (Peter, 2008). Evolutionary transitions between reward and deception may not be uncommon in orchids, as there have been reports of both rewarding and deceptive species in a number of orchid genera (van der Cingel, 1995, 2001; Johnson et al., 1998). Uncovering the evolutionary basis for these transitions is therefore a major challenge for plant reproductive biology.
Nectar is expected to increase not only overall visitation rates, with pollinators visiting more flowers in a patch compared to deceptive species, but also overall PTE, as pollinators should show greater foraging constancy. Indeed, overall pollination success and PTE in the rewarding species A. cochlearis is much higher than in the deceptive A. capensis (Table 1 and Fig. 3B) and in the exclusively deceptive species of Eulophia examined to date. Although PTE in the A. cochlearis population exceeded 60 % in some weeks, the average for 13 Eulophia species surveyed ranged from 11 % to 28 % (Peter, 2008).
A more comprehensive comparison of PTE between the rewarding and deceptive species of Acrolophia will be enlightening. Utilizing the survey of rewarding and deceptive species listed by van der Cingel (1995, 2001), it is possible to identify at least 14 genera containing both rewarding and deceptive species. Direct comparisons of PTE among rewarding and deceptive species, however, could be confounded by other factors, such as pollen vector. For this reason it will probably be most profitable to either search for rewarding and deceptive sister taxa with common pollinators or to attempt to experimentally add nectar to whole populations of deceptive taxa.
The evidence presented in this study indicates that A. cochlearis has a highly specialized pollination system involving a single bee species in the genus Colletes. These short-tongued bees feed on the concentrated nectar with a sucrose to hexose ratio of 2·33. Baker and Baker (1990) surveyed over 700 species and found that short-tongued bees prefer hexose-rich nectars. However, Schmidt-Lebuhn et al. (2007) found that nectars of bee-pollinated species in Acanthaceae are concentrated and dominated by sucrose. Similarly, Petanidou (2005) found that nectars of bee-pollinated Mediterranean species across a number of families are dominated by sucrose. These contrasting results suggest that sugar composition is not a particularly critical aspect for bee pollination.
To date, little attention has been given to the biology of Acrolophia besides curiosity over the unusual ‘black’ flowers of A. ustulata (Kurze and Kurze Hilde, 1990). Russell (2005) observed high rates of fruit set in A. cochlearis in the Western Cape and suggested that rain-assisted autogamy might be at play in this species as has been suggested for Oeceoclades maculata (González-Díaz and Ackerman, 1988). However, from the data presented here, high rates of capsule production seem to be a consequence of high rates of visitation by bees. Because Acrolophia species do not have brightly coloured flowers, it is likely that future investigations will show that fragrance plays a key role in attraction of bees.
The combination of nectar rewards with large numbers of flowers on plants of A. cochlearis (and A. micrantha) must greatly increase the risk of geitonogamy. Acrolophia cochlearis and A. micrantha both have a pollinarium reconfiguration mechanism and, at least in A. cochlearis, there is evidence that this reconfiguration can partly protect the species from geitonogamy as reconfiguration takes longer than the average visits observed (see also Peter and Johnson, 2006b).
The amassing of pollinia that was observed on many of the captured bees (Fig. 1D) could have important implications for the efficiency of the pollinarium reconfiguration system, as well as the likelihood of pollination itself. On the other hand, should a pollinator bearing a large number of pollinaria (e.g. up to 40 pollinia) visit a plant, the odds of geitonogamy would be reduced simply because freshly removed self-pollinia are in a minority among the many pollinaria carried. The relatively open architecture of A. cochlearis flowers allows the short-tongued bees to probe flowers in variable positions such that most of the pollinia on the surface of the half sphere of the pollen mass have an equal chance of being deposited on the stigma of the flower.
Finally, the rapid senescence of visited flowers (Table 2) implies that, in many instances at the height of flowering, there are relatively few flowers open at any one time and often only one fresh flower open per inflorescence branch. This phenomenon reduces the potential for geitonogamy (Harder and Johnson, 2005), but only when visitation rates are high. Earlier in the season, when visitation rates are lower (Fig. 3) there are more virgin flowers open on an inflorescence at one time (e.g. Fig. 1A, B), and geitonogamy might then be more prevalent. The finding that flowers rapidly senesce following removal of pollinaria differs from previous studies of orchids that showed that deposition of pollinia, but not pollinarium removal, has a strong effect on senescence (Proctor and Harder, 1995; Luyt and Johnson, 2001). Senescence of flowers following removal of a pollinarium could result in loss of female function and suggests strong selective pressure against having a large number of flowers open simultaneously in order to minimize pollen discounting.
Direct assessment of rates of self-pollination in A. cochlearis would be desirable. Ideally this would take the form of direct pollinia tracking using histochemical or powder stains (Peakall, 1989; Kropf and Renner, 2008) or microtags (Nilsson et al., 1992), but given the small size of pollinia and presence of an anther cap that completely encloses the pollinia neither of these two established methods is feasible in this species. An alternative approach to estimating rates of selfing and outcrossing entails germinating seeds and using allozyme markers to estimate the degree of inbreeding (Eckert and Barrett, 1994; Ortiz-Barney and Ackerman, 1999). However, this technique is slow, costly and unreliable in the case of orchids that require sterile germination of seeds on nutrient media (Thompson et al., 2001).
Given the difficulties in making direct measurements of self-pollination in A. cochlearis, an alternative method based on fruit quality was used. The different ratios of fertile to infertile seeds produced by self- and cross-pollinated capsules (Table 3) allow the parentage of naturally produced fruit to be determined with a high degree of confidence. This method does not involve manipulation of flowers and is much easier than other techniques to implement. Differences between self- and cross-fertilized fruits in the proportion of seeds containing embryos have been shown in many studies of orchid species (Tremblay et al., 2005; Jersáková et al., 2006). The basis for this is not well understood, but it is believed to result from severe inbreeding depression at the embryo development stage.
Variation in genetic load among populations may limit the applicability of the method to the same populations in which experimental cross- and self-pollinations were performed. Thus, the inference that levels of self-pollination were much higher in a small population of A. cochlearis because of the distribution of fruit quality in that population (Fig 4) should be considered tentative because the effect of self-pollination on the percentage of seeds with embryos was not quantified in that population.
The technique only works when selfed and out-crossed fruits clearly differ in their proportion of seeds with embryos, a situation that is clearly not met in all orchids (cf. Tremblay et al., 2005; Jersáková et al., 2006). It is also obviously not suited to orchids with massulate pollinia where stigmas can receive a mixture of self- and cross-pollen. Interspecific pollinia movement may also obscure the signal, being confused with either self- or out-crossed pollination (cf. Table 3; Kallunki, 1981; Borba et al., 2001), although this is less of a problem in species with highly specialized pollination systems. Despite these limitations, this technique should be applicable to many epidendroid orchids and, thus, to a large proportion of the family.
The NFR and Rhodes University are thanked for providing funding, Ben Eric van Wyk for analysis of the sugar composition of the nectar, Fred Gess for initial identification of insects and Michael Kuhlmann for identifying the pollinator species. Pan and Darwin provided the motivation for the weekly PTE surveys. Grateful thanks to Greig Russell who suggested improvements to the manuscript.