| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Reproductive Biology |
Department of Biological Sciences, Simon Fraser University, 8888 University Drive, Burnaby, British Columbia V5A 1S6 Canada
Received for publication October 8, 2002. Accepted for publication January 16, 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: autogamy • Collinsia parviflora • floral morphology • plant mating systems • pollinator limitation • reproductive assurance • Scrophulariaceae • Vancouver Island
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Charlesworth and Charlesworth, 1987
; Uyenoyama et al., 1993
). Inbreeding depression will reduce the automatic transmission advantage (Fisher, 1941
) of plants with mixed mating systems but lacking for obligate outcrossers. The automatic transmission advantage occurs because mixed mating plants pass on their genes not only through outcrossed pollen and ovules, but also through selfed pollen. More recently, research and models have focused on ecological mechanisms that oppose the evolution of selfing, including pollen discounting, when pollen used for selfing reduces the amount of pollen available for outcrossing (Holsinger et al., 1984
), and seed discounting, when selfed ovules are unavailable for fertilization by outcrossed pollen (Lloyd, 1992
). On the other hand, selfing should be favored if it increases reproductive assurance, the formation of seeds via autonomous selfing when the potential for outcrossing is reduced or absent. Reproductive assurance (RA) selfing is thus expected to be especially important in marginal habitats or when pollinators are rare (Stebbins, 1957
).
The RA hypothesis has lately received much attention. Numerous studies have documented the ability of plants to self when grown in the absence of pollinators (Dole, 1990
; Herrera et al., 2001
; Elle and Hare, 2002
) or have invoked the hypothesis to explain the propensity of selfing phenotypes at the margins of species ranges or under conditions of pollen limitation (Motten, 1982
; Piper et al., 1986
; Charlesworth and Charlesworth, 1987
; Cruden and Lyon, 1989
; Fausto et al., 2001
). Seed production when pollinators are excluded demonstrates the ability of plants to self autonomously, but not its importance or frequency in natural populations. Greater pollen limitation in obligate outcrossers relative to selfers, as determined by pollen supplementation experiments, suggests that selfers derive a benefit through RA, but a lack of difference in pollen limitation does not demonstrate the absence of an RA benefit. For example, full seed set may occur in outcrossers via pollination and in selfers via autonomy. Thus, experimental manipulations in which the ability to self-pollinate autonomously is removed are more appropriate tests of RA than pollen supplementation experiments (Schoen and Lloyd, 1992
) and have rarely been done (but see Kephart et al., 1999
; Herlihy and Eckert, 2002
). In addition, although the importance of RA provided by selfing is expected to be habitat-specific (Cruden and Lyon, 1989
; Schoen et al., 1996
), experimental investigations of RA have only rarely been carried out under natural pollination conditions (Eckert and Schaefer, 1998
; Herlihy and Eckert, 2002
) and never within a single species displaying significant morphological variation for traits expected to influence the mating system. The latter is important because using a single, variable species allows us to stringently test how morphology may be related to any RA benefits accrued without the complication of phylogenetic non-independence.
One obvious trait expected to affect the mating system is flower size. Selfing taxa often have reduced flower size when compared to related outcrossing taxa (Ritland and Ritland, 1989
; Dole, 1992
; Schoen et al., 1997
; Eckhart and Geber, 1999
). Reduced flower size may evolve because of allocation trade-offs and selection for conservation of resources that are no longer needed for the attraction of pollinators in selfing taxa (Charnov, 1982
). An alternative hypothesis is that selection for rapid development in ephemeral habitats leads to reductions in flower size, which subsequently increases the selfing rate (Guerrant, 1989
; Hill et al., 1992
; Aarssen, 2000
). Regardless of the specific mechanism, reduced flower size can increase the ease of autonomous pollen transfer if sexual parts are in contact at anthesis or if spatial proximity increases the effectiveness of delayed selfing mechanisms (Dole, 1990
; Kalisz et al., 1999
). We predict that the importance of RA selfing should vary with flower size, such that small-flowered individuals or taxa should have greater RA through selfing than large-flowered individuals or taxa.
The benefit of RA selfing may also be context dependent. Pollinators are notoriously variable in space and time (e.g., Thompson, 2001
). Abiotic factors like temperature and precipitation can limit pollinator distribution and activity (Cruden and Lyon, 1989
), and different pollinator species vary in their seasonal patterns of abundance (O'Toole and Raw, 1991
). In addition, the co-flowering plant community will affect foraging decisions of the extant pollinator community and may lead to interspecific competition for pollinators (Waser, 1983
). Together, these factors affect the "pollination environment," which could in turn affect the mating system of the plant species of interest. If small-flowered taxa evolved under pollen limitation (Stebbins, 1957
) because of aspects of plant phenology, the environment, or the presence of competitors, the RA benefit of selfing may be especially large for small-flowered plants in the experimental context of a small-flowered background population. Large-flowered plants may not have well-developed autonomous selfing mechanisms, and when placed into a small-flowered background population may have little, if any, seed set through outcrossing if pollinators are rare or absent. In contrast, neither large- nor small-flowered plants may exhibit an RA benefit of selfing in the experimental context of a large-flowered background population, if pollinators are common in such habitats and visit both flower types equally. Of course, the flower types may not be visited equally in experimental populations if pollinator visitation patterns are influenced by flower size.
Here, we examine differences in the potential for RA through autonomous selfing for plants from populations of Collinsia parviflora. We present data on interpopulation differences in flower size and ask: Do pollinators differ among plant populations with different flower sizes? Do pollinators discriminate among plants that differ in flower size? Do plants that differ in flower size differ in the level of RA gained by selfing under different pollination environments? We predict lower pollinator visitation rates in small-flowered populations, pollinator discrimination against small-flowered plants, and greater RA for small flowered plants, especially in small-flowered populations if pollinators are rare as predicted.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Armbruster et al., 2002
); Collinsia parviflora, the small-flowered species, is common from British Columbia south to California, and east to Ontario and Pennsylvania. Collinsia grandiflora is more restricted in distribution, primarily found west of the Cascades in B.C. and only infrequently south to California and Utah (Douglas et al., 1998
). This taxonomy is in dispute in B.C., however. Although in California these species are morphologically distinct and diploid (Garber, 1956
, 1958
; Armbruster et al., 2002
), in the Pacific Northwest populations are morphologically variable and tetraploid (Ganders and Krause, 1986
), suggesting an allotetraploid complex (S. Armbruster, Norwegian University of Science and Technology, personal communication). Ganders and Krause (1986)
recommend recognizing a single species (C. parviflora) as a result of their finding of complete interfertility among B.C. populations that differ in flower size. Two subspecies were distinguished to recognize differences in flower size (subsp. parviflora and subsp. grandiflora; Ganders and Krause, 1986
). However, variation among populations is continuous (see later), which brings each of these taxonomic treatments into question. A more in-depth systematic treatment of Collinsia in B.C. is certainly needed. For the purposes of this paper, we are assuming that sampled populations belong to a single taxon, as in Ganders and Krause (1986)
, and consider the continuous variation in flower size to be of great utility for examination of relationships between variation in flower size and both pollinator choice and RA.
On Vancouver Island, B.C., C. parviflora blooms from March through June (depending on location). Flowers have two upper banner petals, two lower wing petals, and the sexual parts are contained within a folded keel petal. The petals unite at the mouth of the flower to form a corolla tube, which is mildly to strongly saccate on the upper side, forming (in some populations) a pronounced bend. The small nectary is located on the upper side of the base of the ovary, near the saccate bend in the corolla. There are four anthers, two long and two short. Differences in flower size have a quantitative genetic basis in C. parviflora (Ganders and Krause, 1986
; E. Elle, unpublished data). The fruit is a capsule; when ripe, the red seeds are dropped to the ground.
Population survey
Flower size was surveyed in populations of C. parviflora on Vancouver Island, B.C., in 2000 and 2001 (Fig. 1). Appropriate habitat (rock outcrops and deep soil sites within Garry Oak [Quercus garryana] woodlands, beaches, small oceanic islets, and rock jetties) is patchy in distribution and occurs within a dense matrix of forest dominated by Douglas-fir (Pseudotsuga menziesii) and Western redcedar (Thuja plicata). In 2000, two 10-m transects were placed approximately 5 m apart through the center of each of 10 populations. The width across the two attached banner petals was measured on one flower of the plant closest to the tape at 0.5 m intervals, for a total sample size of 40 flowers per population. In 2001, this survey was repeated in seven of the original 10 populations and in two additional populations. Three 10-m transects were laid as described, and one flower was sampled at 1-m intervals, for a total of 30 flowers per population. In addition to flower width, in 2001 we measured the length of the floral tube from where the banner and wing petals unite to the bend in the saccate corolla tube. Because populations differ in the degree of bend in the corolla tube (A. Parachnowitsch and E. Elle, Simon Fraser University, unpublished data), measurements of the length of the tube from the mouth of the flower to the base of the ovary would have differed in accuracy among populations. For seven of the nine populations censused in 2001, each measured flower was collected, placed in ethanol, and brought back to the laboratory where flowers were dissected and ovule number was counted. In all populations, we counted the number of open flowers on each plant that had a flower measured and counted the number of plants present in 400-cm2 quadrats laid along transects at 1-m intervals (sample size for density = 30 quadrats per population). Densities were transformed into units of number of plants per square meter prior to analysis. Finally, we counted total flower production after flowering had ended in all populations (pedicels are retained in this species) along three transects placed as above; we counted pedicels on the plant closest to the tape at 1-m intervals (total sample size = 30).
|
). Finally, we performed a correlation analysis on the plant traits measured in 2001 (corolla width, corolla length, ovule number, and number of open flowers), using the sequential Bonferroni correction to adjust significance levels for multiple comparisons (Rice, 1989
). Total flower number was not included in this analysis because the data was collected at a different time on different plants.
Pollinator diversity
In 2001, pollinator diversity was surveyed in three populations ("test sites"), Thetis Lake Park (TL, flower width in 2001: 4.90 ± 0.08 mm); Cowichan Garry Oak Preserve (GO, flower width in 2001: 5.02 ± 0.14 mm); and Stoltz Meadow (SM, flower width in 2001: 7.69 ± 0.17 mm; Fig. 1). Pollinator surveys were not meant to be comprehensive or replicates of one another, but rather were intended to document the presence/absence of potential pollinators in at least some plant populations.
The test sites were chosen based primarily on access, but differed in some biologically interesting ways. Thetis Lake Park is a public park with numerous trails and moderate to high levels of disturbance (primarily from trampling). Collinsia parviflora and its associated wildflower community occurs on moss-covered rock outcrops surrounding the lake. Cowichan Garry Oak Preserve is an ecological preserve, closed to the public and so with a low disturbance level. This test site has a well-developed soil layer. Stoltz Meadow occurs on rock outcrops on Crown Land. Although there is public access, trails are minimal and so disturbance via trampling is moderate.
At each site, floral visitors (henceforth called pollinators, although the effectiveness of each species as a pollinator has not been investigated) were surveyed on sunny days in 0.25-m2 quadrats placed haphazardly in areas of high floral density. Our objective was to determine whether potential pollinators were present and visiting C. parviflora and other common plant species, all of which were patchy in distribution, and so quadrats could not be placed randomly. Pollinator surveys were done on the entire co-flowering plant community to allow us to distinguish between lack of pollinators and presence of pollinators but lack of visitors to C. parviflora. A total of eight different quadrat locations were used at each site. Within each quadrat, the number of flowering stems for each plant species present was counted. Quadrats were observed for 15-min intervals at least three times (at approximately 1000 hours, 1200 hours, and 1400 hours) in a given day over several days for a total of 6 h of observation at each site. A floral visitor was considered a pollinator if it landed on a flower and/or was observed to probe a flower with its mouthparts. Pollinators were identified to species when possible, to genus or family when not, and the number of flowering stems (by species) visited by each insect was recorded. Contingency tables were constructed of pollinator species by plant species; because of zeros in the data, Fisher's exact test was computed, with the null hypothesis that visitation rates should be proportional to the number of flowering stems of each visited plant species. Some flowering plant species had no visits recorded in our quadrats, so these species were deleted from the data set prior to analysis. Finally, some of the pollinators were combined into functional groups prior to performing Fisher's exact test (see Results).
Plant propagation
For the pollinator choice and RA experiments (see later), plants were grown from seeds collected in 2000 from the two largest-flowered populations (EF and CR) and the two smallest-flowered populations (NH and TL) censused in that year. Seeds from 30 to 40 individuals were pooled from each of these "source sites" and germinated in sterile potting mix. Plants were grown to flowering size in a Conviron growth chamber at Simon Fraser University (16 h 20°C day/8 h 10°C night). Because the number of days between germination and first flower correlates positively with flower size in C. parviflora (E. Elle, unpublished data), seeds from EF and CR were planted in February, and seeds from TL and NH were planted in March, in quantities described below for experiments in May (TL, GO) and June (SM) of 2001. At all test sites, experiments were conducted near, or just past, peak flowering of the native population.
Pollinator choice
At each of the same three test sites used to estimate pollinator diversity (TL, GO, SM), preferences of pollinators for large and small flowers were determined. Two hexagonal choice arrays were placed near flowering C. parviflora at each test site. In each array there were six 10-cm pots, and each pot contained either three small-flowered or three large-flowered plants grown to flowering size in a growth chamber as described earlier. The floral display was adjusted for each pot in the morning such that each pot within an array had the same number of open flowers (range: 10–30). Large- and small-flowered pots were in alternate positions in the array. Within each test site, the array was observed for a total of 50 foraging bouts; at different test sites, this took different amounts of time depending on weather, pollinator density, and other uncontrolled variables. Arrays were rotated and moved to new locations daily. A bout was defined as a single pollinator entering the array, visiting one or more flowers, and leaving the array. Because pollinators were not individually marked, it is possible that individual insects visited the array multiple times. We identified the pollinators to genus (to species where possible) and counted the total number of flowers of each size visited in the bout. Bouts ranged from one to 32 flowers visited. Because visits by single pollinators to multiple flowers within bouts were not independent, data from each bout were summarized for analysis as the proportion of visits to large flowers. Within sites, we tested the null hypothesis that the proportion of visits to large flowers over the 50 foraging bouts did not differ from 0.5 using t tests.
Reproductive assurance experiment
At TL (small-flowered test site) and SM (large-flowered test site), we performed an experiment to estimate the potential RA benefit gained by both small- and large-flowered plants. We constructed eight experimental arrays, each of which held 12 pots in an array consisting of five alternating rows of two and three pots. At each test site, four arrays were used, two that paired plants from TL and CR source sites, and two that paired plants from NH and EF source sites. Each pot contained one plant, and each row contained plants of the same flower size; two arrays had small-flowered plants in the two-pot rows, two arrays had small-flowered plants in the three-pot rows. Thus, all arrays had six small- and six large-flowered plants, and each plant had neighbors of both flower sizes. Plants were grown from seed to flowering size in a growth chamber at Simon Fraser University as described, taken to the field for experiments, and then returned to the growth chamber until seeds (if any) were ripe. This design allowed us to carry out a manipulative experiment under natural pollination conditions and yet control both the composition of our experimental arrays and the phenology of individual plants.
Each plant in each array had two replicate flowers in each of two treatments for a total of four flowers used per plant. Intact flowers were marked on the calyx with colored paper correction fluid but otherwise were unmanipulated. Emasculated flowers had anthers removed in bud and were also marked with colored paper correction fluid. Correction fluid colors were alternated between flower-size classes among arrays, in the event that color affected pollinator visitation. Seeds set in emasculated flowers can only be the result of vector-assisted pollen movement; no effort was made to distinguish between outcrossing or geitonogamy in this experiment, and seed set is referred to as outcrossing for simplicity. This simplification was justified because the purpose of this experiment was to examine the relative contribution of autonomous selfing to total seed production. Seeds set in intact flowers could result from either autonomous selfing or vector-assisted pollen movement. The difference between the two treatments indicates the magnitude (and thus the benefit) of autonomous selfing (if any) when intact seed set is greater than emasculated seed set.
The efficacy of our emasculation treatments was examined in growth chamber plants by comparing seed set of intact and emasculated flowers that either were or were not hand-pollinated (N = 12 flowers for each of these four treatments for two flower sizes, 96 flowers total). Emasculated, unpollinated flowers never set any seeds, and within flower size classes (small or large) there was no difference in seed set between intact, hand-pollinated flowers and emasculated, hand-pollinated flowers (determined by ANOVA with treatment, pollination, and their interaction as main effects and Ryan's Q post-hoc test; data not shown). Thus, we have confidence that our emasculation of flowers for the field experiment was both effective and unlikely to have caused damage or reduced seed set.
To analyze the data from the field experiment, we calculated mean seed number per treatment per plant (usually a mean of two replicate flowers). We subsequently performed an ANOVA with test site (SM or TL), treatment (emasculated or intact), flower size (small or large), and all interactions as fixed effects in the model. We treated flower size as a categorical variable in this analysis for simplicity because no intermediate-flowered source sites were included in our design; there was no overlap in flower sizes in our arrays (corolla width [±1 SE] and range for our arrays: large, 7.52 ± 0.08 mm, range 5.93–9.68 mm; small, 3.34 ± 0.03 mm, range 2.63–4.36 mm). Array within site was included as a random effect in the model (arrays differed in the pair of source sites that provided seeds), thus we utilized PROC MIXED in SAS for this analysis, which calculates standard errors more appropriately than PROC GLM in the analysis of mixed-model ANOVAs (SAS Institute, 1996
). Degrees of freedom for the mixed-model ANOVA were estimated using the Satterthwaite method. Because one of our small-flowered seed sources (TL) was the same as our small-flowered test site, we performed an additional analysis to test for local adaptation. We performed an ANOVA on small-flowered plants only, with seed number as the response variable and test site, treatment, source site, and all interactions as effects in the model. We interpret a significant source site effect or interactions between source site, treatment, and test site as evidence of local adaptation of TL genotypes.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
. Flower size was negatively correlated with ovule number and positively correlated with the number of open flowers per plant (Table 2).
|
|
15 species of forbs were in flower at the time of our pollinator survey at each site, the majority of insect visits in our quadrats were to just a few species, including C. parviflora, Plectritis congesta (Valerianaceae), Camassia quamash (Liliaceae), Lomatium utriculatum (Apiaceae), Lithophragma parviflorum (Saxifragaceae), and Cerastium arvense (Caryophyllaceae). With the exception of P. congesta, which was not present at GO, all plant species were present at each site. In our eight haphazardly placed quadrats, the total number of flowering stems of C. parviflora (and total stems of all visited plant species combined) was 243 (670) at TL, 217 (316) at GO, and 339 (740) at SM. The total number of visits observed over the 6-h period was 144 at TL, 170 at GO, and 317 at SM, with 41.2%, 42.1%, and 51.8%, respectively, to C. parviflora.
|
At TL, the most common floral visitor to C. parviflora was Osmia 1. At GO, all three Osmia morphospecies were common visitors to C. parviflora, but Bombus bifarius also visited flowers. At SM, Bombus flavifrons was the most frequent visitor (Fig. 2). We computed Fisher's exact test to determine whether pollinators differentially visited plants, using the proportion of flowering stems of each visited plant species as our null model. For this analysis, the three Bombus species were included separately but the three Osmia species were combined, as were all other pollinators (Apis + hover flies + other), to reduce the contingency tables to a size amenable to analysis. At SM, C. quamash and L. parviflorum were combined into one category as a total of only seven visits to these species was observed. In all cases, visitation of the different pollinator species to the different plant species was significantly nonrandom (chi-square value and probability from Fisher's exact test: TL, 
2 = 103.99, P < 0.00001; GO, 2 = 122.63, P < 0.00001; SM, 2 = 298.90, P < 0.00001; Fig. 2).
Pollinator choice
Pollinators discriminated among different-sized flowers in our choice arrays, with the average proportion of large flowers visited in 50 foraging bouts always greater than the null expectation of 50%. At TL, the proportion of large flowers visited in 50 bouts was 88.8 ± 3.3% (mean ± 1 SE; t49 = 11.69, P < 0.0001); at GO, the proportion was 81.3 ± 4.1% (t49 = 7.67, P < 0.0001); and at SM, the average proportion was 96.7 ± 1.6% (t49 = 29.24, P < 0.0001). Pollinators visited from one to 32 flowers in the array during a bout (TL range 1–16, 3.80 ± 0.50 flowers; GO range 1–32, 5.38 ± 0.86 flowers; SM range 1–16, 3.94 ± 0.52 flowers).
Reproductive assurance experiment
Small-flowered plants have greater RA than large-flowered plants, especially at the large-flowered test site (Table 3, Fig. 3). Small-flowered plants produced more seeds on average from intact flowers than large-flowered plants (Fig. 3), reflecting differences among populations in the number of ovules available for fertilization (Table 1). Separate from this difference in seed number per fruit, the difference between seed set of intact and emasculated flowers was greater for small- than large-flowered plants (treatment x size interaction, Table 3), meaning a greater RA benefit for small-flowered plants relative to large-flowered plants. Small-flowered plants set proportionately more seeds in emasculated flowers when the experiment was performed in a small-flowered test site; in contrast, large-flowered plants tended to produce proportionately more seeds in emasculated flowers when the experiment was performed in a large-flowered test site (size x treatment x test site interaction; Fig. 3). Greater seed set of emasculated small flowers at the small-flowered test site was not due to local adaptation by experimental plants from the TL source site used at the TL test site, which would be indicated if seed set by TL plants was greater than NH plants at the TL test site but not the SM test site (ANOVA on small plants only, source site effect and all interactions with source site nonsignificant with P 0.13). There is apparently greater potential for outcrossing when experimental plants were morphologically similar to plants in the natural population at the test site.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). Pollen limitation may occur when pollinators are rare or absent, or, if present, when pollinators are inefficient at pollen transfer or discriminate against a particular plant species. We predicted that fewer pollinators would be present and visiting plants in small-flowered relative to large-flowered populations. We found that C. parviflora was visited at all sites, although by different pollinator species; the identity of floral visitors to C. parviflora appeared to depend on both flower size (small-bodied Osmia visited small-flowered plants, large-bodied Bombus visited large-flowered plants) and the co-flowering plant community (a switch of B. bifarius to C. parviflora when P. congesta was absent at GO, Fig. 2). Interspecific competition for pollinators has been shown to affect fruit set in several plant species (reviewed in Waser, 1983
) and may lead to selection on aspects of flower phenology or morphology (Campbell, 1985
; Caruso, 1999
, 2000
). The suggestion that C. parviflora and P. congesta compete for pollination by Bombus is certainly worthy of further study, especially if the difference in body size between our Bombus and Osmia species determines their effectiveness as pollinators.
A lack of pollinators at range margins or in ecosystems where putatively selfing phenotypes occur is often assumed. For example, habitat differences have been invoked to explain variation in the outcrossing rate in several species (Stebbins, 1957
; Vasek, 1964
; Arroyo, 1973
; Schoen, 1982
; Holtsford and Ellstrand, 1992
). An empirical approach was taken in one recent study, which showed that two subspecies of Clarkia xantiana that differ in flower size, selfing rate, and geographic distribution also differed in the available pollinator community (Fausto et al., 2001
). Fewer pollinators were observed visiting small-flowered, selfing populations of C. xantiana subsp. parviflora compared to large-flowered, outcrossing populations of C. xantiana subsp. xantiana, and fewer Clarkia bees (specialist pollinators) were present within the geographic range of selfing C. xantiana subsp. parviflora. In our study, pollinators were present in all sites, but visitation rates were lower in two small-flowered populations than one large-flowered population. Visitation rates are expected to correlate positively with plant density (Fausto, et al., 2001
), and C. parviflora stem number was greater in our observation quadrats at SM than at TL or GO. However density measured along transects (a less subjective measure of population density) indicate high variability and no significant differences among populations, with a trend towards higher density at TL and GO (small-flowered sites) than SM (large-flowered site, Table 1). This suggests that density cannot completely explain differences in visitation rate in C. parviflora. Whether visitation differences were solely due to differences in attractiveness between small- and large-flowered plants or due to differences in the pollinator community among sites that were unrelated to flower size cannot be determined without studying more sites over more flowering seasons.
Pollinators demonstrated a preference for large flowers over small flowers in our experimental arrays. Pollinators have frequently been shown to prefer large flowers or larger floral displays (Bell, 1985
; Conner, 1997
) potentially because flower size advertises floral reward. We controlled the size of the floral display in our pollinator choice experiment, but our natural population survey indicates a positive correlation between flower size and display size in C. parviflora. This result is opposite of theoretical predictions of a trade-off between flower size and number (references in Worley and Barrett, 2000
) and may reflect greater resource acquisition ability in large-flowered plants (van Noordwijk and de Jong, 1986
). If pollinators respond to variation in flower number as well as variation in flower size, large-flowered plants may have an even greater advantage over small-flowered plants than we document here. Large flowers should evolve due to pollinator choice in all populations unless there is some selective benefit to producing small flowers, such as their higher ovule number, or an RA benefit for small-flowered plants in at least some environments or years.
As we predicted, small-flowered C. parviflora produce more seeds through autonomous selfing than large-flowered plants under natural pollination conditions. This RA benefit was especially pronounced in the large-flowered test site, where intact small flowers produced 11.55 times the seeds of emasculated flowers. In the small-flowered test site, intact small flowers produced 3.18 times the seeds of emasculated flowers, a smaller RA benefit than in the large-flowered test site. This smaller benefit was contrary to our prediction that RA should be especially beneficial in the small-flowered test site and may be because one key assumption behind the RA hypothesis, that RA and associated phenotypes occur in habitats where C. parviflora pollinators are rare or absent (Stebbins, 1957
), is overly simplistic. Visitation rates to C. parviflora were indeed lower at our small-flowered test site, but the most frequent visitors were small-bodied Osmia spp., which may be more effective pollinators of small flowers than the large-bodied Bombus spp. that were most common at the large-flowered test site. Collinsia parviflora pollen is placed on the abdomen of the bee when her mass pushes open the folded keel petal and exposes the sex parts. Pollen placement likely differs between large and small flowers and between large and small bees. In addition, although visitation rates were higher at the large-flowered test site, discrimination against small-flowered plants was especially strong, making the effective pollen limitation of small-flowered plants especially high.
Emasculated and intact flowers of large-flowered plants had similar seed set: intact large flowers produced 1.11 times the seeds of emasculated flowers in the large-flowered test site and 1.61 times more seeds than emasculated flowers in the small-flowered test site. This indication of a slight RA benefit of selfing for large-flowered plants in the small-flowered test site could be expected because of the lower visitation rate per flowering stem at this test site relative to the large-flowered test site or if small-bodied bees are less effective pollinators of large flowers.
Whether the RA selfing documented here is beneficial for small-flowered C. parviflora depends on other potential costs and benefits. One unmeasured factor is the extent of inbreeding depression in this species; inbreeding depression would counteract the fitness benefit of increased seed production via autonomous selfing (Lande and Schemske, 1985
; Charlesworth and Charlesworth, 1987
; Uyenoyama et al., 1993
). Because C. parviflora is an annual plant, however, the production of some, albeit inbred, progeny is better than the production of no offspring. The same cannot be said for perennial plants that may receive long-term fitness benefits if resources are used for maintenance rather than inbreeding if future opportunities for outcrossing exist (Lloyd, 1992
; Morgan et al., 1997
).
A second factor that may be important for C. parviflora is the amount of time available for reproduction. The time limitation hypothesis suggests that small flowers, with close proximity between sexual parts and resulting high autonomous selfing rates, evolved in ephemeral habitats where the time available between germination and seed set is limited (Guerrant, 1989
; Hill et al., 1992
; Aarssen, 2000
). In these habitats, there may not be time to develop a large flower and wait for a pollinator before seed maturation can commence. Small-flowered populations of C. parviflora tend to occur in drier parts of the species' range in British Columbia (mean annual precipitation at William Head, near TL, is 911.7 mm; at the Cowichan Lake Forestry Center, near SM, it is 2163.4 mm; Environment Canada climate normals 1971–2000; http://www.msc-smc.ec.gc.ca/climate/), and although both sites are characterized by pronounced summer drought, the drought begins earlier at TL than SM (months with <70 mm precipitation at William Head, April–September; at Cowichan Lake Forestry Center, June–September; Environment Canada). Small-flowered C. parviflora reach reproductive maturity about 1 mo earlier than large-flowered plants in the common environment of the growth chamber (E. Elle, unpublished data), possibly the result of selection for rapid development. We purposefully removed this potential selective force from our experiment by using potted plants with staggered planting times, but limited time for reproduction in particular habitats will lead to greater benefits of autonomous selfing in those habitats. An increased benefit for selfing would make the level of inbreeding depression (e.g., the cost) required for selection to act against selfing prohibitively high.
Variation in flower size and the mating system is likely under complex selection in C. parviflora. The present study illustrates one benefit of small flower size, an increase in reproductive assurance via autonomous selfing under natural pollination conditions. It also illustrates a potential cost, in that pollinators discriminate against small flowers when given a choice, which will be important if there is inbreeding depression or pollen or seed discounting in this species. Future research will explore these costs, as well as the importance of climate and development time for mating system evolution, and how spatial and temporal variation in the pollinator community affects the selective framework for flower size in this species.
|
FOOTNOTES |
|---|
2 Author for reprint requests (phone: 604-291-4592; Fax: 604-291-3496; eelle{at}sfu.ca
)
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Armbruster W. S. C. P. H. Mulder B. G. Baldwin S. Kalisz B. Wessa H. Nute 2002 Comparative analysis of late floral development and mating-system evolution in tribe Collinsieae (Scrophulariaceae S.L). American Journal of Botany 89: 37-49
Arroyo M. T. K. 1973 A taximetric study of intraspecific variation in autogamous Limnanthes floccosa (Limnanthaceae). Brittonia 25: 177-191[CrossRef][ISI]
Bell G. 1985 On the function of flowers. Proceedings of the Royal Society of London 224: 223-265
Campbell D. R. 1985 Pollinator sharing and seed set of Stellaria pubera: competition for pollination. Ecology 66: 544-553[CrossRef][ISI]
Caruso C. M. 1999 Pollination of Ipomopsis aggregata (Polemoniaceae): effects of intra- vs. interspecific competition. . American Journal of Botany 86: 663-668
Caruso C. M. 2000 Competition for pollination influences selection on floral traits of Ipomopsis aggregata. Evolution 54: 1546-1557[ISI][Medline]
Charlesworth D. B. Charlesworth 1987 Inbreeding depression and its evolutionary consequences. Annual Review of Ecology and Systematics 18: 237-268[CrossRef][ISI]
Charnov E. L. 1982 The theory of sex allocation. Princeton University Press, Princeton, New Jersey, USA
Conner J. 1997 Floral evolution in wild radish: the roles of pollinators, natural selection, and genetic correlations among traits. International Journal of Plant Science 158: S108-S120[CrossRef][ISI]
Cruden R. W. D. L. Lyon 1989 Facultative xenogamy: examination of a mixed mating system. In J. H. Bock and Y. B. Linhart [eds.], The evolutionary ecology of plants, 171–207. Westview Press, Boulder, Colorado, USA
Day R. W. G. P. Quinn 1989 Comparisons of treatments after an analysis of variance in ecology. Ecological Monographs 59: 433-463[CrossRef][ISI]
Dole J. A. 1990 Role of corolla abscission in delayed self-pollination of Mimulus guttatus (Scrophulariaceae). American Journal of Botany 77: 1505-1507[CrossRef][ISI]
Dole J. A. 1992 Reproductive assurance mechanisms in three taxa of the Mimulus guttatus complex (Scrophulariaceae). American Journal of Botany 79: 650-659[CrossRef][ISI]
Douglas G. W. D. V. Meidinger J. Pojar 1998 Illustrated flora of British Columbia. Ministry of Environment, Lands, and Parks, Victoria, British Columbia, Canada
Eckert C. G. A. Schaefer 1998 Does self-pollination provide reproductive assurance in Aquilegia canadensis (Ranunculaceae)?. American Journal of Botany 85: 919-924[Abstract]
Eckhart V. M. M. A. Geber 1999 Character variation and geographic distribution of Clarkia xantiana A.Gray (Onagraceae): flowers and phenology distinguish two subspecies. Madroño 46: 117-125
Elle E. J. D. Hare 2002 Environmentally induced variation in floral traits affects the mating system in Datura wrightii. Functional Ecology 16: 79-88[CrossRef][ISI]
Fausto J. A. J. V. M. Eckhart M. A. Geber 2001 Reproductive assurance and the evolutionary ecology of self-pollination in Clarkia xantiana (Onagraceae). American Journal of Botany 88: 1794-1800
Fisher R. A. 1941 Average excess and average effect of a gene substitution. Annals of Eugenics 11: 53-63[ISI]
Ganders F. R. G. R. Krause 1986 Systematics of Collinsia parviflora and C. grandiflora (Scrophulariaceae). Madroño 33: 63-70
Garber E. D. 1956 The genus Collinsia. I. Chromosome number and chiasma frequency of species in the two sections. Botanical Gazette 118: 71-73[CrossRef]
Garber E. D. 1958 The genus Collinsia. VII. Additional chromosome numbers and chiasmata frequencies. Botanical Gazette 120: 55-56[CrossRef]
Guerrant E. O. J. 1989 Early maturity, small flowers, and autogamy: a developmental connection?. In J.H. Bock and Y. B. Linhart [eds.], The evolutionary ecology of plants, 61–84. Westview Press, Boulder, Colorado, USA
Herlihy C. R. C. G. Eckert 2002 Genetic cost of reproductive assurance in a self-fertilizing plant. Nature 416: 320-323[CrossRef][Medline]
Herrera C. M. A. M. Sanchez-Lafuente M. Medrano J. Guitian X. Cerda P. Rey 2001 Geographical variation in autonomous self-pollination levels unrelated to pollinator service in Helleborus foetidus (Ranunculaceae). American Journal of Botany 88: 1025-1032
Hill J. P. E. M. Lord R. G. Shaw 1992 Morphological and growth rate differences among outcrossing and self-pollinating races of Arenaria uniflora (Caryophyllaceae). Journal of Evolutionary Biology 5: 559-573[CrossRef][ISI]
Holsinger K. E. M. W. Feldman F. B. Christiansen 1984 The evolution of self-fertilization in plants: a population genetic model. American Naturalist 124: 446-453[CrossRef][ISI]
Holtsford T. P. N. C. Ellstrand 1992 Genetic and environmental variation in floral traits affecting outcrossing rate in Clarkia tembloriensis (Onagraceae). Evolution 46: 216-225[CrossRef][ISI]
Kalisz S. D. Vogler B. Fails M. Finer E. Shepard T. Herman R. Gonzales 1999 The mechanism of delayed selfing in Collinsia verna (Scrophulariaceae). American Journal of Botany 86: 1239-1247
Kephart S. R. E. Brown J. Hall 1999 Inbreeding depression and partial selfing: evolutionary implications of mixed-mating in a coastal endemic, Silene douglasii var. oraria (Caryophyllaceae). Heredity 82: 543-554
Lande R. D. W. Schemske 1985 The evolution of self-fertilization and inbreeding depression in plants. I. Genetic models. Evolution 39: 24-40[CrossRef][ISI]
Lloyd D. G. 1992 Self- and cross-fertilization in plants. II. The selection of self-fertilization. International Journal of Plant Science 153: 370-380[CrossRef]
Morgan M. T. D. J. Schoen T. M. Bataillon 1997 The evolution of self-fertilization in perennials. American Naturalist 150: 618-638[CrossRef][ISI]
Motten A. F. 1982 Autogamy and competition for pollinators in Hepatica americana (Ranunculaceae). American Journal of Botany 69: 1296-1305[CrossRef][ISI]
O'Toole C. A. Raw 1991 Bees of the world. Blandford, London, UK
Piper J. G. B. Charlesworth D. Charlesworth 1986 Breeding system evolution in Primula vulgaris and the role of reproductive assurance. Heredity 56: 207-217[ISI]
Rice W. R. 1989 Analyzing tables of statistical tests. Evolution 43: 223-225[CrossRef][ISI]
Ritland C. K. Ritland 1989 Variation of sex allocation among eight taxa of the Mimulus guttatus species complex (Scrophulariaceae). American Journal of Botany 76: 1731-1739[CrossRef][ISI]
SAS Institute. 1996 SAS/STAT software: changes and enhancements for release 6.12. SAS Institute, Cary, North Carolina, USA
Schoen D. J. 1982 The breeding system of Gilia achilleifolia: variation in floral characteristics and outcrossing rate. Evolution 36: 352-360[CrossRef][ISI]
Schoen D. J. M. O. Johnston A.-M. L'Heureux J. V. Marsolais 1997 Evolutionary history of the mating system in Amsinckia (Boraginaceae). Evolution 51: 1090-1099[CrossRef][ISI]
Schoen D. J. D. G. Lloyd 1992 Self- and cross-fertilization in plants. III. Methods for studying modes and functional aspects of self-fertilization. International Journal of Plant Science 153: 381-393[CrossRef][ISI]
Schoen D. J. M. T. Morgan T. M. Bataillon 1996 How does self-pollination evolve? Inferences from floral ecology and molecular genetic variation. Philosophical Transactions of the Royal Society of London 351: 1281-1290[CrossRef]
Stebbins G. L. 1957 Self fertilization and population variability in the higher plants. American Naturalist 91: 337-354[CrossRef][ISI]
Thompson J. D. 2001 How do visitation patterns vary among pollinators in relation to floral display and floral design in a generalist pollination system?. Oecologia 126: 386-394[CrossRef][ISI]
Uyenoyama M. K. K. E. Holsinger D. M. Waller 1993 Ecological and genetic factors directing the evolution of self-fertilization. Oxford Surveys in Evolutionary Biology 9: 327-381
van Noordwijk A. J. G. de Jong 1986 Acquisition and allocation of resources: their influence on variation in life history tactics. American Naturalist 128: 137-142[CrossRef][ISI]
Vasek F. C. 1964 The evolution of Clarkia unguiculata derivatives adapted to relatively xeric environments. Evolution 18: 26-42[CrossRef][ISI]
Waser N. M. 1983 Competition for pollination and floral character differences among sympatric plant species: a review of evidence. In C. E. Jones and R. J. Little [eds.], Handbook of experimental pollination biology, 277–293. Van Nostrand Reinhold, New York, New York, USA
Worley A. C. S. C. H. Barrett 2000 Evolution of floral display in Eichhornia paniculata (Pontederiaceae): direct and correlated responses to selection on flower size and number. Evolution 54: 1533-1545[ISI][Medline]
This article has been cited by other articles:
|
|
 |
|
  C. W Weekley and A. Brothers Failure of reproductive assurance in the chasmogamous flowers of Polygala lewtonii (Polygalaceae), an endangered sandhill herb Am. J. Botany, February 1, 2006; 93(2): 245 - 253. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. T. Ivey and D. E. Carr Effects of herbivory and inbreeding on the pollinators and mating system of Mimulus guttatus (Phrymaceae) Am. J. Botany, October 1, 2005; 92(10): 1641 - 1649. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. W. Busch The evolution of self-compatibility in geographically peripheral populations of Leavenworthia alabamica (Brassicaceae) Am. J. Botany, September 1, 2005; 92(9): 1503 - 1512. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. R. Herlihy and C. G. Eckert Evolution of self-fertilization at geographical range margins? A comparison of demographic, floral, and mating system variables in central vs. peripheral populations of Aquilegia canadensis (Ranunculaceae) Am. J. Botany, April 1, 2005; 92(4): 744 - 751. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. L. Parachnowitsch and E. Elle Variation in sex allocation and male-female trade-offs in six populations of Collinsia parviflora (Scrophulariaceae s.l.) Am. J. Botany, August 1, 2004; 91(8): 1200 - 1207. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||
; focal populations, +; weather stations, 
(see Discussion). Abbreviations follow 
level of 0.05 after sequential Bonferroni adjustment
