| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Botany, University of Toronto, 25 Willcocks Street, Toronto, Ontario, Canada, M5S 3B2
Received for publication April 24, 1998. Accepted for publication October 8, 1998.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: bumble bee behavior • floral color change • Melastomataceae • phenology • pollen limitation • Rhexia virginica • specialization
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). Pollen removal requires bees that land on the flowers and vibrate their indirect flight muscles at a high frequency, which causes pollen to stream out of the anthers. In reference to the sound made by the bee vibrations, this pollination system is called buzz pollination. Recent studies consider the mechanics of pollen removal from poricidal anthers (King and Lengoc, 1993
; King and Ferguson, 1994
) and its implications for pollen dispersal (Harder and Barclay, 1994
; King and Buchmann, 1996
). Nonetheless, field investigations of the reproductive ecology of buzz-pollinated species are few (Macior, 1964
; Buchmann, Jones, and Colin, 1977
; Cane and Payne, 1988
; Knudsen and Olesen, 1993
), and several questions concerning the response of pollinators to floral traits associated with buzz pollination remain unanswered.
The Melastomataceae, a large predominately tropical family characterized by buzz pollination, exemplify two aspects of the buzz pollination syndrome that have received little empirical consideration (Buchmann, 1983
; Renner, 1989
). First, anthers in the family are typically bright yellow in color, which mimics abundant pollen even when they are empty. This has been interpreted as a deceptive adaptation because insects must visit the flowers to determine whether they are rewarding (Vogel, 1978
; Buchmann, 1983
). Second, melastome flowers commonly undergo a color change after anthesis (Renner, 1989
; Weiss, 1995
). Floral color change in flowering plants has been hypothesized to increase the efficiency of pollinator foraging and pollen transfer (Gori, 1983
, 1989
), and the maintenance of flowers that have undergone a color change is thought to increase floral display size and thus visitation rates (Cruzan, Neal, and Willson, 1988
; Gori, 1989
; Weiss, 1991
; but see Casper and La Pine, 1984
; Delph and Lively, 1989
). Here, we investigate the function of these two aspects of the buzz pollination syndrome in the melastome Rhexia virginica L. (Virginia meadow-beauty), using experimental manipulations and observations of bee foraging behavior.
Rhexia virginica is a perennial wetland herb found predominately on the coastal plain of the United States, but following glacial retreat it migrated inland to the vicinity of the Great Lakes (Reznicek, 1994
; Fig. 1A). Populations in the Muskoka region of Ontario, Canada, are significantly disjunct in the distribution of the species and are at the edge of the familial range of the Melastomataceae. In this region, the floral syndrome of R. virginica is unique because the few co-occurring buzz-pollinated taxa have either a solanoid or ericoid floral morphology. In contrast, the nectarless flowers of R. virginica consist of four large, showy pink petals and eight elongate, bright yellow, poricidally dehiscent anthers that spread laterally from the center of the flower. The style is sigmoidal and directed downwards so that the stigma is below the anthers. Given the distinctiveness of R. virginica in Ontario, an investigation of its pollination syndrome provided an opportunity to assess its function in a geographic context that is marginal with respect to the family's distribution.
|
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). At least 50 local populations of R. virginica are scattered along its shoreline (Keddy and Sharp, 1989
), which represents one of the largest concentrations of the species in Canada. We conducted research in a total of eight populations of R. virginica at Lake Matchedash in 1996 and 1997 (Fig. 1). We undertook detailed investigations of the ecology of pollen limitation in populations B and D in 1997 (Larson and Barrett, in press
).
Floral biology of Rhexia virginica
Mating system
The compatibility status of R. virginica is uncertain. Kral and Bostick (1969)
and Sharp (1983)
demonstrated that fruits are set after self-pollination. Based on the former study, however, Renner (1989)
claimed that R. virginica is self-incompatible because seeds from selfed flowers did not germinate. To determine whether R. virginica flowers from Lake Matchedash set seed after self-pollination and whether they depend on insect visits for pollination to occur, we performed an experimental pollination study in a glasshouse at the University of Toronto in July and August 1996. Each treatment included a minimum of 15 plants transplanted from populations B, D, G, H, and I at Lake Matchedash. We assigned flowers randomly to one of four treatments: self-pollination, cross-pollination, unpollinated to test for autonomous self-fertilization, or "disturbed." We obtained pollen for pollination treatments by slitting anthers and using forceps to remove pollen that was subsequently applied to stigmas. We watered "disturbed" flowers from above with a watering can for 4 min to simulate whether rain could potentially cause self-pollination (see Renner, 1989
).
The pollen : ovule ratio of angiosperm species provides some indication of their mating system (Cruden, 1977
). To calculate the pollen : ovule ratio of R. virginica, we counted ovules in 25 flowers sampled from each of the four terminal nodes on plants in population D at Lake Matchedash on 10 August 1996. We also collected anthers from a total of 154 flowers during 1996 and placed them in 75% ethanol in separate microcentrifuge tubes. We released pollen grains from the anthers using a probe sonicator and counted them using a particle counter (see Harder, 1990
, for details).
Floral color change
After flowering for 1 d R. virginica flowers undergo a color change (Weiss, 1995
). To determine whether pollinator visits or environmental cues induce this color change (Gori, 1983
), we qualitatively compared the color of 30 flowers located in pollinator exclosure cages at Lake Matchedash to flowers outside exclosures. To assess whether flowers having undergone a color change (second-day flowers) could contribute directly to plant fertility if visited by insects, we compared the receptivity of second-day stigmas and viability of second-day pollen to that of first-day flowers using controlled hand-pollinations with outcross pollen. Treatments were applied to plants located within exclosure cages in population D during early August 1996.
Flowering phenology
The number of flowers within plant populations varies through time, and pollinator densities may be associated with the distribution of flowering (Thomson, 1980
). We characterized the phenology of R. virginica at Lake Matchedash in 1996 by randomly choosing 50 plants along shoreline transects within populations A, B, C, and D prior to flowering. We surveyed plants every other day until flowering ceased and recorded the anthesis date of each flower. We also recorded second-day flowers to quantify their contribution to floral display size.
Pollination biology of Rhexia virginica
Floral visitation
To identify the pollinators of R. virginica flowers, we watched visitors during observation periods throughout peak flowering in 1996 and 1997 at Lake Matchedash. We quantified rates of floral visitation in 1996 by recording the number of visitors entering single 4-m2 quadrats in populations B and D every hour from 0700 until 1500 for 15-min periods on 10 d during the flowering period.
Bumble bee behavior
We undertook four investigations of bumble bee foraging behavior on R. virginica flowers at Lake Matchedash in 1996 and 1997. (1) We recorded bumble bee visits to second-day flowers to determine whether they were visited in proportion to their abundance or whether the bees preferentially visited first-day flowers. (2) To determine whether bumble bees had a preference for unvisited R. virginica flowers compared to previously visited flowers, we presented bees with experimental arrays of 16 first-day flowers on 4 and 14 August 1996. We positioned flowers in florist water pics located 5 cm from one another in a square array comprising equal numbers of flowers of two randomly arranged treatments: (a) flowers that had been visited during the morning and from which we expelled remaining pollen by repeated tapping using forceps and (b) unvisited flowers from an exclosure cage. (3) To quantify the proportion of pollen that can be removed from R. virginica flowers without "buzzing," we conducted an investigation prior to bumble bee visitation on 10 August 1997 in population D. We tapped single flowers from ten plants repeatedly with forceps until we could remove no more pollen and then estimated the amount removed by comparison with the amount in unmanipulated flowers. (4) Bumble bees could cause self-pollination in R. virginica by transferring pollen (i) within flowers during visits (facilitated self-pollination sensu Lloyd and Schoen, 1992
) or (ii) between flowers on a plant (geitonogamous self-pollination). We assessed the potential for bumble bees to transfer pollen geitonogamously, within inflorescences, by recording the number of R. virginica flowers they visited during foraging on multiflowered inflorescences in 1996.
Female fertility of Rhexia virginica
Survey of patterns of fruit set
We conducted investigations in 1996 and 1997 to assay fruit set in R. virginica and to determine the extent to which it varied in space and time. We measured fruit set in populations A, B, C, and D at Lake Matchedash in 1996 by monitoring the 200 plants marked for the phenology study. We measured fruit set in 1997 at seven populations at Lake Matchedash and in single populations from Krapek Lake (79°48' W, 45°13' N), Three-mile Lake (79°16' W, 44°53' N), Bentshoe Lake (78°55' W, 45°02' N), Morrison Lake (79°28' W, 44°52' N), and Hardy Lake (79°32' W, 45°00' N) (Fig. 1). We counted the number of buds on 25 randomly chosen plants in each population in early August and recorded the presence of fruits on these plants in October. We also estimated the size of populations visited in 1997 to determine whether there was a relation between female fertility and population size.
Factors affecting female fertility
We investigated the effect of various factors on the female fertility of R. virginica in populations A, B, C, and D at Lake Matchedash in 1996. We marked flowers on plants in the phenology study individually with paint so that fruit set and the number of seeds produced could be related to the population in which a plant was located, flowering date, and the number of flowers displayed on the date of flowering (daily display size). We treated flowering date categorically as early, middle, or late in the season, corresponding to flowering of one-third of all flowers counted on plants in the phenology study. To assess whether second-day flowers increased pollinator visitation rates and hence fertility, we either included or excluded them in display size and compared the results. We analyzed fruit set using logistic regression, with population, display size, and flowering date treated as independent variables. Since fertility levels in the multiple flowers comprising some daily displays were not independent, we used the proportion of these flowers setting fruit in the analysis. In 94% of daily displays, plants set either all fruit or no fruit. In the remaining displays we considered fruit set complete or zero in the logistic regression, depending on whether or not the majority (>50%) of fruits set. To determine the effect of the same variables on seed set, we conducted a mixed-model ANOVA. All statistical analyses were conducted using JMP (Version 3.0.2, SAS, 1994
).
Pollen limitation of fertility
We undertook an investigation at Lake Matchedash in 1996 to determine whether the female fertility of R. virginica was pollen limited. We randomly selected between 15 and 30 plants with two flowers in populations B and D on 1, 5, 8, and 17 August. The two populations were of similar size and contained 
1000 plants. We added supplemental pollen from a nearby plant to one flower and compared its fruit set and seeds per fruit to the control flower. We conducted all pollination treatments on clear, sunny days unless otherwise noted. Preliminary investigations indicated that pollen limitation occurred at the whole-plant level (sensu Johnston, 1991
) in R. virginica at Lake Matchedash, so confounding of treatment effects with resource re-allocation within inflorescences was unlikely (B. M. H. Larson and S. C. H. Barrett, unpublished data).
To assess whether pollen limitation limited female fertility of R. virginica flowers elsewhere in the Muskoka region, we conducted surveys at Bentshoe, Hardy, Krapek, and Three-mile Lakes (Fig. 1) in 1997. We visited each lake once during the period 7–14 August, selected pairs of one-flowered plants of similar size, and added supplemental pollen to the stigma of one flower per pair. We selected plants with a floral display size of one because this was the modal daily flower number (see Results).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
± 1 SE = 140.4 ± 7.9 seeds; self = 142.4 ± 11.1; t166 = 0.15, NS). Neither flowers that were not experimentally hand-pollinated (N = 51) nor those subjected to simulated rain (N = 31) set fruit.
The pollen : ovule ratio of R. virginica flowers at Lake Matchedash was 668. Flowers produced many small pollen grains ( ± SE = 3.05 x 105 ± 5.63 x 103 grains, N = 154; ± SE = 20.47 ± 0.09 µm, N = 200) and numerous ovules ( ± SE = 456.6 ± 7.5 ovules, N = 87). Comparison of the quantity of pollen produced by each of the eight anthers in a flower revealed no significant difference among them ( ± SE = 3.2 x 104 ± 8.5 x 103 grains, F7, 95 = 0.65, NS). Similarly, ovule counts did not differ among flowers from different nodes on the plant (F3, 83 = 1.73, NS).
Floral color change
Floral color change within exclosures at Lake Matchedash differed little from that outside exclosures, indicating that it was not induced by visitation. Color change was most marked in the filaments, which became red and recurved on the second day. Reddening of anthers in second-day flowers was slight, but their yellow color was not apparent in a frontal view of the flower because they were hidden by the curved filaments. Temperature affected the rate of visible color change; on warm days filaments were dark red and fully recurved by late afternoon, whereas during cool periods they often remained pale reddish and only slightly recurved until the next morning. The size of first- and second-day flowers was similar, but second-day petals were paler and slightly smaller than first-day petals (first-day petal length, ± SE = 11.65 ± 0.14 cm; second-day = 11.42 ± 0.15 cm; N = 48 pairs, Ts = 182.5, P < 0.06, Wilcoxon's signed-ranks test). Petals occasionally remained attached to the flower for 2 d after anthesis.
Controlled hand-pollinations at Lake Matchedash demonstrated that the female and male fertilities of second-day flowers were both low. Second-day flowers had significantly lower fruit set compared to first-day flowers whether we pollinated them with pollen from first-day (fruit set = 11.8%, N = 17) or second-day flowers (fruit set = 6.3%, N = 16) (df = 3, 
2 = 41.84, P < 0.0001, G test of independence). Fruit set was lower when we pollinated first-day stigmas with second-day pollen than first-day pollen (using first-day pollen, fruit set = 100%, N = 13; using second-day pollen, fruit set = 57.1%, N = 14; df = 1, 2 = 9.48, P < 0.005), but the number of seeds produced per fruit was similar (first-day pollen, ± SE = 76.1 ± 16.0 seeds; second-day pollen = 53.4 ± 15.9; t19 = 0.72, NS, square-root transformed data).
Flowering phenology
Phenological investigations of R. virginica at Lake Matchedash indicated that its flowering season lasted from mid-July to early October. However, the vast majority of flowering occurred from late July to mid-August. Peak flowering was in early August, and the distribution of flowering over the season was not markedly skewed (Fig. 2).
|
± SE = 5.3 ± 0.3, range = 1–18, N = 197; Fig. 3A), which bloomed over a 9-d period. On days that plants flowered they displayed a median of one first-day flower ( ± SE = 1.4 ± 0.03 flowers, range = 1–5, N = 735; Fig. 3B). Second-day flowers infrequently augmented the size of displays consisting of first-day flowers: 74.6% of the daily floral displays that contained first-day flowers did not contain second-day flowers. Therefore, the median total daily display size was only slightly greater than display sizes that included first-day flowers (median = 1 flower, ± SE = 1.8 ± 0.05, range = 1–8; Fig. 3B). The largest daily floral display size we observed consisted of seven first- and four second-day flowers.
|
0630 EST) and typically declined in early afternoon. Analysis of quadrat data indicated that bumble bees made 82% of visits (N = 97) to flowers in 1996. The majority (75%) of these visits were by Bombus bimaculatus, with declining proportions attributable to B. terricola, B. ternarius, and B. affinis, respectively. Each of these species was infrequent in 1997. Instead, B. impatiens was the predominant visitor and made >90% of recorded floral visits (N = 65). Visitation rates during 1996 were quite variable. The median number of visits to quadrats during 15-min observation periods was one ( ± SE = 1.4 ± 0.17 visits, N = 59). The majority of the time we recorded no visits, but there were occasionally as many as five visits to quadrats (Fig. 4).
|
3 s ( ± SE = 3.13 ± 0.10 s, range = 0.6–13.95 s, N = 410). While buzzing, they arched their body beneath them and contacted the anthers and often the stigma. In this position, buzzing expelled pollen onto both their thorax and abdominal tergites, imparting their abdomen with a characteristic white tip. Halictid bees spent lengthy periods on flowers ( ± SE = 16.96 ± 2.92 s, range = 1–120 s, N = 53) and buzzed individual anthers, but they were relatively uncommon and made little contact with the stigma, so they were probably minor pollinators.
Bumble bee behavior
During observation periods at Lake Matchedash in 1996 and 1997 bumble bees occasionally approached second-day flowers, but rarely visited them. For example, on 9 and 10 August 1997, the frequency of second-day flowers in population B was 46 and 42%, respectively. Despite this, only eight (0.3%) of 2341 bumble bee visits recorded on these two days were to second-day flowers.
Array experiments at Lake Matchedash in 1996 demonstrated that bumble bees were more likely to visit unvisited R. virginica flowers than those that had been previously visited. On both days bumble bees made more visits to previously unvisited flowers located in the arrays (visits to unvisited flowers = 78; visits to previously visited flowers = 36; Gpooled = 15.84, df = 1, P < 0.0001), but the strength of this pattern differed between days (Gheterogeneity = 4.00, df = 1, P < 0.05).
Experiments in 1997 demonstrated that buzz pollination was not the only mechanism that released pollen grains from R. virginica anthers. We removed a significant proportion (49.3%) of pollen by repeated tapping of the anthers (pollen count prior to manipulation, ± SE = 2.96 x 105 ± 1.10 x 104 grains; after manipulation = 1.50 x 105 ± 5.57 x 103, t22 = 12.02, P < 0.0001, log-transformed data).
Observation of bumble bee foraging indicated that their behavior could result in geitonogamous pollen transfer. Bumble bees showed a tendency to visit an increasing proportion of flowers on larger inflorescences than smaller ones (39.5, 53.5, and 63.2% of visits were to more than one flower on inflorescences with two, three, and four first-day flowers, respectively; N = 367, 2 = 10.52, df = 2, P < 0.01, G test of independence).
Female fertility of Rhexia virginica
Survey of patterns of fruit set
Population surveys of R. virginica in the Muskoka region of Ontario in 1996 and 1997 revealed three clear patterns in female fertility. First, mean fruit set of plants in the populations was consistently below maximum (percentage fruit set, ± SE = 52.6 ± 0.02%, N = 530). In 1996, fruit set among the four populations studied at Lake Matchedash was 60.6% (N = 195, population range = 40.4–72.6%), and among the 12 populations surveyed in the Muskoka region in 1997 it was 47.9% (N = 335, population range = 5.7–74.2%). Second, in both years levels of fruit set were highly variable among populations (1996: F3, 191 = 8.09, P < 0.0001; 1997: F11, 323 = 10.62, P < 0.0001, arcsine square-root transformed data). Lastly, in the 12 Muskoka populations in 1997, we found a significant relation between fruit set and size (Fig. 5). Taken together, these patterns highlight the low and variable fertility of R. virginica populations in Muskoka. This was further demonstrated by annual variation in fruit set. We assessed fertility in 1996 and 1997 in populations B and D at Lake Matchedash. An ANOVA indicated that the fertility of these two populations was markedly different in the two years, with only the interaction between year and population significant (F1, 199 = 19.92, P < 0.0001).
|
2 = 30.86, df = 2, P < 0.001, G test of independence). The logistic model accounted for a relatively small proportion of variation in fruit set (r2 = 10.4%), indicating that additional ecological factors must also be important. The only factor that contributed significantly to the three-way ANOVA of seed set in flowers setting fruit was display size (first-day flowers: F1, 431 = 5.56, P < 0.05; total display: F1, 431 = 4.10, P < 0.05, square-root transformed data).
|
2 = 37.24, df = 1, P < 0.0001, G test of independence), as was mean number of seeds per fruit (control, ± SE = 67.76 ± 7.58 seeds; supplemented = 115.75 ± 6.89 seeds; N = 55 pairs, Ts = 472.5, Pone-tailed < 0.001, Wilcoxon's signed-ranks test). The fertility of open-pollinated controls was particularly low near the beginning of the flowering season, probably as a result of wet, overcast weather.
|
± SE = 97.87 ± 11.31 seeds, N = 15; supplemented = 138.88 ± 8.28, N = 15; Wilcoxon Z = 2.57, P < 0.01). This variation in pollen limitation was not directly associated with the number of flowers displayed within populations (Fig. 7).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Floral biology of Rhexia virginica
Flowers of R. virginica depend on insects for pollen transfer because fruits were not set autonomously or when we subjected flowers to simulated rain. Contrary to the statement by Renner (1989)
, flowers were self-compatible, and the occurrence of self-pollination is suggested by the pollen : ovule ratio of R. virginica and observations of pollinator behavior. Bumble bees were the predominant pollinators of R. virginica at Lake Matchedash, and they may cause self-fertilization in two ways. They could transfer pollen between flowers on a plant (geitonogamous self-fertilization), but the relative paucity of multiple daily flowers on inflorescences at Lake Matchedash (Fig. 3B) implies that this contributes minimally to rates of selfing. A greater source of self-fertilization at Lake Matchedash is likely to be intrafloral facilitated self-fertilization caused during bumble bee visits. This may occur directly if pollen that lands on their bodies while they buzz is later deposited on the stigma. Alternatively, pollen from the cloud they disperse as they buzz may land on the stigma. Further investigations are required to determine the prevalence of these modes of self-fertilization in R. virginica and the implications of self-fertilization for seed fitness. Selfed seeds from populations of R. virginica in North Carolina did not germinate (Kral and Bostick, 1969
), but this has not been examined for populations elsewhere in the range.
This study highlights the various perspectives from which floral phenology can be viewed. At a population level the flowering distribution of R. virginica was not markedly skewed (Fig. 2). Positively skewed flowering distributions are common and may be favored when a rapid initiation of flowering counteracts the initial hesitancy of pollinators to visit novel floral resources (Thomson, 1980
). If the floral syndrome of a species is similar to species previously flowering in the community, less skewed distributions may result. This is unlikely to account for the nonskewed phenology of R. virginica, because its flowers were quite dissimilar to all other species found at Lake Matchedash. It is possible that its flowering distribution reflects the lag time required for bumble bees to sample it as a "minor" species and to learn to buzz pollinate, before adopting it as a "major" species (Heinrich, 1976a
). This hypothesis could be explored by determining the "innateness" of buzzing by bumble bees and the amount of experience required before they efficiently acquire pollen from R. virginica flowers (Laverty, 1994
).
The anthesis of flowers on plants of R. virginica was quite dispersed, with typically only one flower in anthesis during a single day (Fig. 3B). This results in an extended flowering period, with few individual flowers in anthesis simultaneously, as has been noted for buzz-pollinated species in general (Buchmann, 1983
) and for R. virginica in particular (Leggett, 1881
). The apportioning of pollen into numerous flowers functions as a pollen-packaging mechanism that may reduce the diminishing returns associated with the removal of a large proportion of a plant's pollen during single pollinator visits (Harder and Thomson, 1989
). A lengthy flowering period also increases the probability that some flowers set seed, even if occasional poor weather precludes some flowers from fruiting.
Function of floral color change
As reported for other flowers that undergo color change, second-day flowers of R. virginica were infertile and not visited by pollinators (Weiss, 1991
, 1995
). One potential function of these flowers is to increase floral display size, which attracts greater numbers of pollinators to the inflorescence (reviewed in Klinkhamer and de Jong, 1990
) and, hence, increases the probability of fruit set. The maintenance of second-day flowers generally increased the size of floral displays of R. virginica at Lake Matchedash (Fig. 3B), and their contribution to fertility was demonstrated by the greater ability of floral display size to predict fruit set in R. virginica when it included second-day flowers (Table 1). Nonetheless, most inflorescences at Lake Matchedash had only one daily flower and second-day flowers contributed infrequently to display size. Second-day flowers may enhance floral display size more markedly in the center of diversity for the genus Rhexia, in the southeastern United States, where R. virginica plants are larger and have more flowers (James, 1956
; S. C. H. Barrett, personal observations). Adaptive explanations for floral color change and the maintenance of second-day flowers may apply more to these populations, rather than those at the edge of the range where a shortened growing season appears to limit plant size and hence flower production.
The maintenance of sterile flowers contributes to floral display size, but what is the function of concurrent color change? Without color change, visitors would likely visit both first- and second-day flowers on inflorescences. Previous studies have recognized that this may result in pollen wastage if pollen from one plant is transferred to sterile flowers on another (Gori, 1983
; Cruzan, Neal, and Willson, 1988
). However, these studies have not explicitly considered the wastage of pollen resulting from transfer between first- and second-day flowers within an inflorescence. Pollen transferred in this way is unavailable for deposition on stigmas of other plants (pollen discounting sensu Harder and Barrett, 1995
). Increased visitation associated with larger displays would therefore be counterbalanced by reductions in outcrossed male fertility. More pollen discounting with larger displays seems likely in R. virginica because pollinators visited a larger proportion of flowers as more were displayed. With color change, however, pollinators avoid sterile flowers. Collectively, our results are consistent with the hypothesis proposed by Harder and Barrett (1996)
that the adaptiveness of floral color change is that it increases insect visitation rates by contributing to daily display size, without the associated mating costs resulting from pollen discounting. This hypothesis could be tested by analyses of the costs and benefits of variation in floral display size in different parts of the range of R. virginica.
Floral color change modifies the cues provided by petals and anthers to foraging pollinators. Petals generally act as long-distance attractants to foraging bumble bees, but it may be the contrast between anthers and petals that determines whether a visit occurs (Lunau, 1990
, 1996
). Preliminary choice experiments with R. virginica demonstrated reduced visitation to flowers from which either petals or anthers had been removed, suggesting that visitation depends upon cues provided by both of these structures (B. M. H. Larson and S. C. H. Barrett, unpublished data). In R. virginica, floral color change was localized to the androecium. This is relatively uncommon among flowering plants, occurring in only 14% (N = 77) of families for which floral color change has been documented (Weiss, 1995
). Nonetheless, in species where the sole reward is pollen, the status of anthers often provides a short-distance cue of whether flowers are rewarding (Harder and Barclay, 1994
; Weiss, 1995
). The yellow anthers of R. virginica did not appear to be deceptive to visiting bumble bees, because they were less likely to approach flowers that had previously been visited than virgin flowers (see also Zimmerman, 1982
; Cresswell and Robertson, 1994
). It is uncertain whether the signal to bumble bees was the lower pollen content of the anthers, small red necrotic marks ("bee kisses") left where bees held the anthers while buzzing (e.g., Renner, 1989
) or scent markings (e.g., Goulson, Hawson, and Stout, 1998
).
Is buzz pollination in Rhexia virginica specialized?
Buzz pollination may be considered specialized if only one or a few bees act as pollinators. However, unlike flowers offering nectar rewards, which partition the bumble bee fauna according to proboscis length (Heinrich, 1976b
; Harder, 1985
), buzz-pollinated flowers offering pollen rewards could potentially be rewarding to all bumble bee species in a given region. In the case of R. virginica at Lake Matchedash, a variety of bumble bee species were pollinators, and there was a switch from Bombus bimaculatus as the predominant pollinator in 1996 to B. impatiens in 1997. The varied pollinators of R. virginica imply that there is little partitioning of the local bumble bee fauna. Generalized visitation by multiple bumble bee species has also been reported in other buzz-pollinated taxa (e.g., Knudsen and Olesen, 1993
; Harder and Barclay, 1994
).
Pollination systems dependent upon bumble bees are relatively specialized, because bumble bee learning capacity gives them a high capacity for floral constancy (Heinrich, 1976a
). In Ontario populations, however, three observations imply that bumble bee foraging on R. virginica may not be particularly specialized in terms of the efficiency of pollen transfer. First, a large amount of pollen could be removed without buzzing, which contrasts with the statement by Buchmann (1983)
that pollen is very difficult to remove from poricidal anthers without high-frequency vibration. This result indicates that the dispensing function of poricidal anthers (sensu Harder and Thomson, 1989
) may be compromised in R. virginica because pollen can be removed simply by bee movement on the flowers. Second, pollen dispersed as a cloud when bumble bees buzzed, which suggests that pollen deposition on their bodies was not particularly localized. This observation supports Renner's (1989)
hypothesis that deposition is generalized in most melastomes, but quantification of the distribution of pollen on bumble bee bodies would be required to explicitly test this hypothesis. In particular, it is quite unusual for pollen to accumulate on the dorsal surface of the abdomen (Buchmann, 1983
; Renner, 1989
), and the amount present here could be compared to that on the ventral section of the thorax, which is probably exposed to pollinator grooming, as well as to that elsewhere on the body. Lastly, Buchmann (1983)
suggested that little pollen is wasted while bees buzz, but the previous observation suggests otherwise. The "efficiency" of pollen deposition could be assessed by comparisons with taxa related to R. virginica.
The consistent submaximal fertility of R. virginica in Muskoka suggests that bumble bees were unreliable pollinators. The mean fruit set in all populations investigated was 52.6%, which was similar to the 56% recorded at Axe Lake, Ontario in 1982 (N = 25 flowers; Sharp, 1983
). This level of fertility is markedly lower than the average fruit set of 72.5% among 445 self-compatible hermaphroditic species in a survey conducted by Sutherland and Delph (1984)
. There are several adaptive explanations for low fertility in plants (reviewed in Sutherland, 1986
), but the proximate mechanism in Ontario populations of R. virginica appears to be insufficient pollen transfer by pollinators (Figs. 6, 7; Larson and Barrett, in press
). Pollen limitation may have been less in larger populations, accounting for the significant positive relation between total population size and the fruit set of individual plants (Fig. 5). Prevalent pollen limitation of R. virginica in Ontario suggests that the geographic marginality of populations may have compromised the functioning of its pollination syndrome. However, whether consistent pollen limitation has any demographic and fitness consequences for Ontario populations is not clear. Comparative investigations of the pollination ecology and demography of populations in the center of the range in the southeastern United States with those we studied in Ontario would be required to determine whether geographic marginality influences the functioning of the buzz pollination syndrome of R. virginica.
|
|
FOOTNOTES |
|---|
2 Author for correspondence (phone 416 978-4151, Fax 416 978-5878, e-mail: barrett{at}botany.utoronto.ca
).
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
———, C. E. Jones and L. J. Colin. 1977 Vibratile pollination of Solanum douglasii and S. xanti (Solanaceae) in southern California. Wasmann Journal of Biology 35: 1–25.
Cane, J. H., and J. A. Payne. 1988 Foraging ecology of the bee Hapropoda laboriosa (Hymenoptera: Anthophoridae), an oligolege of blueberries (Ericaceae: Vaccinium) in the southeastern United States. Annals of the Entomological Society of America 81: 419–427.
Casper, B. B., and T. R. La Pine. 1984 Changes in corolla color and other floral characteristics in Cryptantha humilis (Boraginaceae): cues to discourage pollinators? Evolution 38: 128–141.[CrossRef][ISI]
Cresswell, J. E., and A. W. Robertson. 1994 Discrimination by pollen-collecting bumblebees among differentially rewarding flowers of an alpine wildflower, Campanula rotundifolia (Campanulaceae). Oikos 69: 304–308.[CrossRef][ISI]
Cruden, R. W. 1977 Pollen-ovule ratios: a conservative indicator of breeding systems in flowering plants. Evolution 31: 32–46.
Cruzan, M. B., P. R. Neal, and M. F. Willson. 1988 Floral display in Phyla incisa: consequences for male and female reproductive success. Evolution 42: 505–515.[CrossRef][ISI]
Delph, L. F., and C. M. Lively. 1989 The evolution of floral color change: pollinator attraction versus physiological constraints in Fuchsia excorticata. Evolution 43: 1252–1262.
Eyde, R. H., and J. A. Teeri. 1967 Floral anatomy of Rhexia virginica (Melastomataceae). Rhodora 69: 163–178.
Gori, D. F. 1983 Post-pollination phenomena and adaptive floral changes. In C. E. Jones and R. J. Little [eds.], Handbook of experimental pollination biology, 31–49. Van Nostrand Reinhold, New York, NY.
———. 1989 Floral color change in Lupinus argenteus (Fabaceae): why should plants advertise the location of unrewarding flowers to pollinators? Evolution 43: 870–881.[CrossRef][ISI]
Goulson, D., S. A. Hawson, and J. C. Stout. 1998 Foraging bumblebees avoid flowers already visited by conspecifics or by other bumblebee species. Animal Behaviour 55: 199–206.[CrossRef][ISI][Medline]
Harder, L. D. 1985 Morphology as a predictor of flower choice by bumble bees. Ecology 66: 198–210.[CrossRef][ISI]
———. 1990 Pollen removal by bumble bees and its implications for pollen dispersal. Ecology 71: 1110–1125.[CrossRef][ISI]
———, and R. M. R. Barclay. 1994 The functional significance of poricidal anthers and buzz pollination: controlled pollen removal from Dodecatheon. Functional Ecology 8: 509–517.
———, and S. C. H. Barrett. 1995 Mating cost of large floral displays in hermaphrodite plants. Nature 373: 512–515.[CrossRef]
———, and ———. 1996 Pollen dispersal and mating patterns in animal-pollinated plants. In D. G. Lloyd, and S. C. H. Barrett [eds.], Floral biology: studies on floral evolution in animal-pollinated plants, 140–190. Chapman and Hall, New York, NY.
———, and J. D. Thomson. 1989 Evolutionary options for maximizing pollen dispersal of animal-pollinated plants. American Naturalist 133: 323–344.[CrossRef][ISI]
Heinrich, B. 1976a The foraging specializations of individual bumblebees. Ecological Monographs 46: 105–128.[CrossRef][ISI]
———. 1976b Resource partitioning among some eusocial insects: bumblebees. Ecology 57: 874–889.[CrossRef][ISI]
James, C. W. 1956 A revision of Rhexia (Melastomataceae). Brittonia 8: 201–230.[CrossRef]
Johnston, M. O. 1991 Pollen limitation of female reproduction in Lobelia cardinalis and L. siphilitica. Ecology 72: 1500–1503.
Keddy, C. J., and M. J. Sharp. 1989 Atlantic Coastal Plain Flora Conservation in Ontario. Report prepared for the Natural Heritage League and World Wildlife Fund.
King, M. J., and S. L. Buchmann. 1996 Sonication dispensing of pollen from Solanum laciniatum flowers. Functional Ecology 10: 449–456.[CrossRef][ISI]
———, and A. M. Ferguson. 1994 Vibratory collection of Actinidia deliciosa (Kiwifruit) pollen. Annals of Botany 74: 479–482.
———, and L. Lengoc. 1993 Vibratory pollen collection dynamics. Transactions of the American Society of Agricultural Engineers 36: 135–140.
Klinkhamer, P. G. L., and T. J. de Jong. 1990 Effects of plant size, plant density and sex differential nectar reward on pollinator visitation in the protandrous Echium vulgare (Boraginaceae). Oikos 57: 399–405.[CrossRef][ISI]
Knudsen, J. T., and J. M. Olesen. 1993 Buzz-pollination and patterns in sexual traits in North European Pyrolaceae. American Journal of Botany 80: 900–913.[CrossRef][ISI]
Kral, R., and P. E. Bostick. 1969 The genus Rhexia (Melastomataceae). Sida 3: 387–440.
Larson, B. M. H., and S. C. H. Barrett. In press The ecology of pollen limitation in buzz-pollinated Rhexia virginica (Melastomataceae). Journal of Ecology.
Laverty, T. M. 1994 Bumble bee learning and flower morphology. Animal Behaviour 47: 531–545.[CrossRef][ISI]
Leggett, W. H. 1881 Fertilization of Rhexia virginica L. Bulletin of the Torrey Botanical Club 8: 102–104.[CrossRef]
Lloyd, D. G., and D. J. Schoen. 1992 Self- and cross-fertilization in plants. I. Functional dimensions. International Journal of Plant Science 153: 358–369.[CrossRef]
Lunau, K. 1990 Color saturation triggers innate reactions to flower signals: flower dummy experiments with bumble bees. Journal of Comparative Physiology A 166: 827–834.
———. 1996 Unidirectionality of floral color changes. Plant Systematics and Evolution 200: 125–140.[CrossRef][ISI]
Macior, L. W. 1964 An experimental study of the floral ecology of Dodecatheon meadia. American Journal of Botany 51: 96–108.
Renner, S. S. 1989 A survey of reproductive biology in neotropical Melastomataceae and Memecylaceae. Annals of the Missouri Botanical Garden 76: 496–518.[CrossRef][ISI]
Reznicek, A. A. 1994 The disjunct coastal plain flora in the Great Lakes region. Biological Conservation 68: 203–215.[CrossRef][ISI]
SAS. 1994 JMP user's guide, version 3.0.2 edition. SAS Institute, Cary, NC.
Sharp, M. J. 1983 Ecological factors affecting the distribution of a rare plant, Rhexia virginica L., at Axe Lake, Ontario. M. Sc. thesis, University of Guelph, Guelph, Ontario.
———, and P. A. Keddy. 1983 Rhexia virginica L. (Melastomataceae). In G. W. Argus, K. M. Pryer, D. J. White, and C. J. Keddy [eds.], Atlas of the rare vascular plants of Ontario. Four parts (looseleaf). National Museum of Natural Sciences, Ottawa, Ontario.
Sutherland, S. 1986 Patterns of fruit-set: what controls fruit-flower ratios in plants? Evolution 40: 117–128.[CrossRef][ISI]
———, and L. F. Delph. 1984 On the importance of male fitness in plants. Ecology 65: 1093–1104.[CrossRef][ISI]
Thomson, J. D. 1980 Skewed flower distributions and pollinator attraction. Ecology 61: 572–579.[CrossRef][ISI]
Vogel, S. 1978 Evolutionary shifts from reward to deception in pollen flowers. In A. J. Richards [ed.], The pollination of flowers by insects, 89–96. Linnean Society Symposium Series No. 6. Academic Press, New York, NY.
Weiss, M. R. 1991 Floral color changes as cues for pollinators. Nature 354: 227–229.[CrossRef][ISI]
———. 1995 Floral color change: a widespread functional convergence. American Journal of Botany 82: 167–185.[CrossRef][ISI]
Zimmerman, M. 1982 Optimal foraging: random movement by pollen collecting bumblebees. Oecologia 53: 394–398.[CrossRef][ISI]
This article has been cited by other articles:
|
|
 |
|
  G. Clausing and S. S. Renner Molecular phylogenetics of Melastomataceae and Memecylaceae: implications for character evolution Am. J. Botany, March 1, 2001; 88(3): 486 - 498. [Abstract] [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||
P < 0.06) and are based on G tests of independence.