| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Reproductive Biology |
2Rocky Mountain Biological Laboratory, Crested Butte, Colorado 81224 USA; 3Department of Biology, Agnes Scott College, Decatur, Georgia 30030 USA; 4Institute of Ecology, University of Georgia, Athens, Georgia 30602 USA
Received for publication January 30, 2003. Accepted for publication May 20, 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: floral morphology • hummingbird pollination • Ipomopsis aggregata • path analysis • Polemoniaceae • pollen receipt • pollinator efficiency • pollinator visitation
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Buchman and Nabhan, 1996
), animal pollination represents an important ecosystem service (Costanza et al., 1997
). Estimating pollinator visitation (i.e., number of visits, percentage of flowers probed per visit), however, to link pollinator visitation to plant reproduction is often time-consuming and challenging (Kearns and Inouye, 1993
). For example, when pollinator visitation is infrequent (Campbell et al., 1991
; Mitchell, 1992
; Ackerman et al., 1997
; Parker, 1997
), a researcher may observe only a handful of pollinator visits during hundreds of observation hours. Moreover, external factors, such as the presence of the researcher, may influence the behavior of floral visitors (akin to the "herbivory uncertainty principle"; Cahill et al., 2001
, 2002
). Unfortunately, we know relatively little about the reliability of indirect estimates of pollinator visitation rates, such as counting mechanically tripped flowers or recording pollination-mediated floral color changes (reviewed in Kearns and Inouye, 1993
).
Here we examine the utility of one indirect measure of pollinator visitation, namely pollen receipt by flowers. We are assuming that any pollen that was deposited on the stigma must have been put there by a pollinator, and increased pollinator visitation should result in increased pollen receipt by stigmas. Although a positive association between pollinator visitation and pollen receipt may hold in some natural systems (e.g., Galen and Stanton, 1989
; Young and Stanton, 1990
; Jones and Reithel, 2001
) or under controlled conditions (Mitchell and Waser, 1992
), pollen receipt as an estimate of pollinator visitation may be misleading if, among other factors, (1) dehiscing anthers directly contact the stigma (reviewed in Barrett, 2002
), (2) stigmas close after pollinator visitation (Fetscher and Kohn, 1999
), (3) pollen receipt is a function of both the number of times flowers are visited by pollinators and pollinator effectiveness at picking up pollen from anthers and depositing pollen on stigmas (Murcia, 1990
; Campbell et al., 1994
; Conner et al., 1995
), (4) pollinators groom pollen from their bodies in successive floral visits (Conner et al., 1995
), (5) pollinators dislodge pollen from the stigma surface (Gori, 1983
; Young, 1988
), or (6) pollen receipt follows a saturating curve because the stigmatic surface area is limited (Galen and Stanton, 1989
). Moreover, if hermaphroditic flowers vary in the proportion of time they spend in the pistillate vs. staminate phase (Campbell, 1989a
; Campbell et al., 1994
), variation in the duration of stigma receptivity may further disrupt the relationship between pollinator visitation and pollen receipt. These factors are not mutually exclusive.
To reduce the probability that pollen receipt is a function of stigma-anther contact or other forms of autogamous pollen transfer, pollinator visitation can be estimated using emasculated flowers. Emasculating flowers eliminates self-pollen deposition. However, functional anthers are necessary for the growth and development of many flowers (Greyson and Tepfer, 1967
), and anther removal may sever hormonal relationships among floral organs and cause developmental abnormalities in flowers (Kiss and Konig, 1990
). Not only does emasculation change floral morphology and development, but it may affect pollinator activity. For example, Rademaker et al. (1997)
found that bumble bees approached emasculated flowers but, compared to unemasculated flowers, landed on the corollas in a "clumsy" manner and usually did not come in contact with the stigma. And when pollen serves as the reward for floral visitors, pollinators may avoid emasculated flowers. In addition, emasculation may reduce levels of geitonogamy (within-plant pollen movement), affecting subsequent plant fitness (e.g., de Jong et al., 1992
).
If there is temporal separation between the maturation of anthers and stigma receptivity (dichogamy) or spatial separation (herkogamy), pollen receipt may provide a reliable estimate of pollinator visitation without emasculation. However, the temporal separation between anther and stigma maturation is often incomplete. Any overlap in reproductive structures could lead to autodeposition of pollen on receptive stigmas (Waser and Price, 1991
), resulting in misleading estimates of pollinator visitation.
Moreover, pollen receipt may not only be a function of the number of times pollinators visit flowers but also a function of pollinator efficiency. We are defining pollinator efficiency as the stigma pollen load deposited per pollinator visit (Inouye et al., 1994
). In this definition, we are ignoring the male component of pollinator efficiency, removing pollen from anthers (Inouye et al., 1994
). Different species of floral visitors often vary in their pollinator efficiency at flowers of the same plant species. For example, Arnold (1982)
found that bumble bees (Bombus spp.) transferred over two times more pollen per visit to flowers of Linaria vulgaris Miller (Scrophulariaceae) than halictid bees (Halictus and Dialictus spp.). In addition, natural phenotypic variation in floral morphology may affect the efficiency of the same pollinator species. For example, Campbell et al. (1994)
found that natural variation in floral morphology (stigma exsertion) influenced hummingbird efficiency at depositing pollen on the stigmas of flowers of Ipomopsis aggregata (Pursh) V. Grant (Polemoniaceae). Given these scenarios, variation in pollinator efficiency because of a diversity of pollinators or because of variation in floral morphology may further affect the relationship between pollinator visitation rate and pollen receipt.
Here we examine the link between pollinator visitation rate and pollen receipt by unemasculated flowers in a naturally growing population of the hummingbird-pollinated Ipomopsis aggregata. Ipomopsis aggregata provides an ideal model plant with which to address the links between pollinator visitation and pollen receipt because I. aggregata is pollinated by a relatively simple pollinator assemblage (primarily two species of hummingbirds). Therefore, variation in pollinator assemblage likely will not impact the relationship between visitation rate and pollen receipt. Quantifying useful estimates of hummingbird pollinator visitation rates to I. aggregata can be challenging as observable pollinator visits to flowers are infrequent. For example, Campbell et al. (1991)
observed average pollinator visitation rates to I. aggregata of only 0.1 visits per flower per hour during peak visitation times (usually dawn and dusk for hummingbird-pollinated species). Therefore, understanding the relationship between pollinator visitation and pollen receipt may provide a useful tool for estimating pollinator visitation in this system.
We hypothesized that increased pollinator visitation would increase pollen receipt across natural variation in floral morphology and pollinator visitation levels. Thus, in addition to quantifying pollinator visitation rate and pollen receipt on flowers of I. aggregata, we measured natural variation in floral and plant characters known to be important to pollinator visitation and/or pollen receipt (Campbell, 1989a
; Wolf and Hainsworth, 1990
). We then dissected the effects of natural variation in floral and plant characters on pollen receipt by looking at pollinator visitation rate vs. pollinator efficiency. Specifically, we asked (1) To what degree does increased pollinator visitation per flower increase pollen deposition? and (2) What is the relative importance of natural variation in floral and plant characters to pollen deposition through changes in pollinator visitation and pollinator efficiency? Finally, because increased pollinator visitation per flower may also result in increased geitonogamy in unemasculated plants, we also counted all of the fruits and seeds produced per plant, and asked (3) Do changes in pollinator visitation and pollen deposition translate into changes in female plant reproductive success (total seeds produced)?
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). Upon bolting, the plant produces a stalk of scarlet, trumpet-shaped flowers (number of flowers produced = 84 ± 66 flowers, mean ± 1 SD; Campbell, 1989b
). Each plant flowers for 4–8 wk. The individual hermaphroditic flowers last for 3–5 d. Flowers are protandrous, but the temporal separation between anthers and stigmas may be incomplete (Waser and Price, 1991
). Ipomopsis aggregata are primarily self-incompatible (Waser and Price, 1991
). Male- and female-phase flowers occur on the same plant at the same time. Flowers produce nectar in both male and female phases, initiating nectar production as buds begin to elongate (Pleasants, 1983
). Individual plants vary in plant and floral morphological characters within and among populations, including variation in plant height, corolla length and width, and floral display size (Campbell, 1989a
, b
). Broad-tailed hummingbirds [Selasphorus platycercus Swainson (Trochilidae)] and rufous hummingbirds (S. rufus Gmelin) are the most common pollinators of I. aggregata around the RMBL (Waser, 1978
).
Field procedures
We randomly marked 40 flowering I. aggregata with similar flowering phenologies on a south-facing slope in a population with approximately 200 flowering I. aggregata. We measured plant height to the nearest 0.1 cm at the onset of flowering and randomly chose one flower per plant and measured corolla length (from the base of the calyx to the floral opening) and corolla width (at the opening of the corolla tube) to the nearest 0.1 mm using digital calipers. Measures of corolla length and width are highly repeatable across flowers within I. aggregata plants (Campbell et al., 1991
; Wolf and Campbell, 1995
); therefore, measuring one flower per plant should provide a reliable estimate of corolla length and width. We also counted the number of open flowers per plant twice a week as an estimate of the size of the floral display. We measured plant height, corolla length and width, and floral display because these characters affect pollinator visitation rate and/or pollinator efficiency in this system (Campbell, 1989a
, b
, 1991
).
From 27 June to 20 July 2001, 5 d per wk, we observed hummingbird foraging behavior from approximately 0530 until 0900, when hummingbirds foraged most actively in this population (E. C. Engel and R. E. Irwin, personal observation). Because the population was relatively small (approximately 10 x 10 m in area), we could observe all study plants at once for pollinator visits. For each observation period, we recorded the number of times each plant was visited and the number of flowers probed per plant visit. We conducted a total of 57 h of observations.
Biweekly, we collected stigmas from 50% of the flowers in the female phase on each plant to quantify the number of pollen grains received as a function of hummingbird visitation. We squashed stigmas in basic fuchsin dye and counted the number of conspecific and heterospecific pollen grains using a compound microscope (Kearns and Inouye, 1993
). We counted pollen grains on a total of 825 stigmas.
Once plants senesced, we collected all of the fruit capsules to assess the relationship between pollen receipt and seed production per plant. We counted the number of seeds in each seed-bearing fruit.
Data analysis
For each plant, we calculated hummingbird visitation rate as (number of times the plant was visited per observation period) x (mean percentage of flowers probed per plant visit per observation period), averaged across the flowering season (square-root transformed). For each plant, this measure represents hummingbird visitation rate on a per-flower basis averaged across observation periods. Because we only observed pollinator visits once per day, hereafter we refer to this variable as visitation rate per observation period or simply visitation rate. We calculated mean pollen receipt per stigma per plant as the mean number of I. aggregata pollen grains received per stigma collected for each plant, averaged across the flowering season. Because very few heterospecific pollen grains were found on the stigmas, we did not include heterospecific pollen receipt in the analyses. We did not compare visitation rate and pollen receipt on a weekly basis because we were primarily interested in the seasonal relationship between pollinator visitation rate and pollen receipt. Initially, we fit linear and quadratic terms for the relationship between visitation rate and pollen receipt. A significant quadratic term might indicate that at the upper bound of visitation rate, pollen receipt would not increase because the stigma is limited in size. We found no evidence for nonlinearity in the relationship and further dissected the linear relationship between visitation rate and pollen receipt below.
We used path analysis (Wright, 1921
, 1934
; Li, 1975
) combined with structural equation modeling (SEM; reviewed in Mitchell, 1992
, 1993
) to examine the fit of four competing a priori hypotheses exploring the relationship between pollinator visitation rate and pollen receipt (Fig. 1). Path analysis allows for the examination of multiple relationships among variables, both direct and indirect, using a set of a priori hypotheses and allows for the estimation and comparison of effect sizes (Kingsolver and Schemske, 1991
; Mitchell, 1992
, 1993
). One caveat about path analysis is that it identifies correlations among variables, not the causal nature of such links (Kingsolver and Schemske, 1991
; Mitchell, 1992
; Petraitis et al., 1996
). In the path diagrams (Fig. 1), the effect of natural variation in floral characters on pollinator efficiency are direct effects on pollen receipt, whereas the effect of natural variation in plant and floral characters on pollinator attraction are indirect effects on pollen receipt mediated through pollinator visitation rate.
|
) and indirectly through changes in attractiveness to pollinators (Campbell et al., 1991
). We include plant height in all models as a proxy for the attractiveness of plants to pollinators. Plant height and mean floral display size are positively and significantly correlated (r = 0.45, N = 40, P = 0.004); therefore, plant height captures both the size of the floral display and the height at which those flowers are advertised to pollinators. Plant height is unlikely to directly influence pollen receipt; therefore, no direct path is drawn between plant height and pollen receipt. However, plant height may have indirect effects on pollen receipt through increased plant attraction to pollinators (Wolf and Hainsworth, 1990
). Finally, pollinator visitation rate directly influences pollen receipt (Mitchell and Waser, 1992
), and plant and floral traits are intercorrelated as a result of both genetic and environmental factors (Campbell, 1996
). Model B is nested within Model A (Fig. 1), with the indirect effects of corolla length and width on pollen receipt mediated through pollinator visitation rate constrained to zero, but all other paths remain identical. We include this second competing model because corolla length and width may have no indirect effects on pollen receipt through changes in pollinator attraction. For example, Campbell et al. (1991)
found that corolla width had no effect on pollinator visitation rate once nectar production and corolla length were controlled for. However, corolla length and width may still have direct effects on pollen receipt through changes in pollinator efficiency. If we find evidence supporting Models A or B, then pollen receipt is dually determined by pollinator visitation rate and pollinator efficiency, suggesting that visitation rate may not be a reliable indicator of pollen receipt alone. Model C is nested within Model A (Fig. 1), with the direct effects of corolla length and width on pollen receipt constrained to zero, but all other paths remain identical. We included this third model because corolla length and width may have no direct effect on pollinator efficiency at depositing pollen but may have indirect effects on pollen receipt mediated through changes in visitation rate. Finally, in Model D (Fig. 1), which is nested within the Models above, both the direct and indirect effects of corolla length and width on pollen receipt are constrained to zero. If we find evidence for Models C or D, this would suggest that visitation rate is directly and solely linked to pollen receipt because corolla length and width have no significant effect on pollinator efficiency at depositing pollen.
We used SEM to test statistically which competing hypothesis, Model A, B, C, or D (Fig. 1), provided the most appropriate fit to the observed data (Hayduk, 1987
; Loehlin, 1987
; Mitchell, 1992
, 1993
). Structural equation modeling tests among models using a goodness-of-fit statistic that approximates a 
2 distribution with df = the difference between the number of observed correlations and the number of coefficients. A nonsignificant 2 value indicates no significant difference between the expected and observed correlation matrices; therefore, the model provides an appropriate fit to the observed data. We also report Akaike's information criterion (AIC); the model that minimizes AIC provides the most parsimonious fit to the data. For the four competing path models, we used PROC CALIS in SAS (METHOD = ML) to calculate AICs, goodness-of-fit statistics, and significance values (SAS, version 8.02). Petraitis et al. (1996)
recommend a sample size of 5–20 times larger than the number of estimated paths in the competing path models. With a sample size of 40 plants (Fig. 1), our sample size is within the acceptable range, albeit on the low side, to conduct SEM. In addition, to screen for collinearity among corolla length, corolla width, and plant height, we calculated variance inflation factors (hereafter referred to as VIFs; VIF option in PROC REG). Correlations among variables inflate VIFs and cause a loss of precision in the path analysis. In all cases, the VIFs were less than 1.1; therefore, it is unlikely that collinearity had a strong impact on the results (Myers, 1990
).
For the path diagram that provided the most appropriate fit to the observed data, we calculated direct and indirect effects and significance values using PROC CALIS. Direct effects represent standardized partial regression coefficients (p, path coefficients). Indirect effects represent the portion of variation jointly determined by the independent and transmitter variables on the dependent variable.
Finally, to ensure that changes in visitation rate and pollen receipt translated into differences in plant reproduction, as has been shown previously for I. aggregata (e.g., Campbell and Halama, 1993
), we examined the relationship between pollen receipt and total seed production per plant. We found no evidence that the relationship between pollen receipt and seed set per plant was nonlinear; we report results from linear regression only using PROC REG.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
|
). We found a nonsignificant difference between Models A and B (2 = 0.5, df = 2, P = 0.8). Because Model B was simpler than Model A and had a lower AIC value than Model A, we retained Model B based on the principle of parsimony and because the model that minimizes AIC provides the most reliable fit to the observed data (Table 2). We then compared Model B to Model D and found a marginally significant decrease in fit for Model D compared to Model B (2 = 5.5, df = 2, P = 0.05). In addition, Model B had a lower AIC value than Model D, suggesting that Model B provided the most reliable fit to the observed data.
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
That we found a positive link between visitation rate and pollen receipt suggests that a variety of mechanisms associated with autodeposition of pollen (i.e., temporal overlap within flowers between pollen dehiscing anthers and receptive stigmas) did not overwhelm the positive relationship. Moreover, increased visitation rate and pollen receipt were also associated with increased seed production. Because I. aggregata is self-incompatible, this increase in seed production suggests that the majority of pollen deposited on stigmas was outcrossed pollen deposited by pollinators and was not self-pollen deposition. The relationship between visitation rate and pollen receipt was linear, suggesting that pollen receipt did not show a decelerating relationship with increased pollinator visitation (Galen and Stanton, 1989
). Visitation rate to I. aggregata flowers is low (Campbell et al., 1991
); therefore, pollinator visitation rates and subsequent pollen deposition levels may only rarely be high enough to result in stigma saturation. Whether the strong relationship between visitation rate and pollen receipt is common in other hummingbird-pollinated plant systems with low pollinator visitation rates remains to be tested. For plant–pollinator systems with high pollinator visitation rates and/or diverse pollinator assemblages, the relationship between number of floral visits and pollen receipt may not have such clear patterns.
We hypothesized that the link between pollinator visitation and pollen receipt may not hold if pollen receipt was also a function of pollinator efficiency, especially for species such as I. aggregata in which floral morphology varies. Our results only partially supported this hypothesis. Model B (Fig. 2) provided the most appropriate fit to the observed data, suggesting that pollen receipt was jointly determined by pollinator visitation rate and pollinator efficiency. In this case, pollinator efficiency was measured as the direct effect of floral morphology on pollen receipt. However, corolla length and width had only weak, nonsignificant effects on pollen receipt, indicating that natural variation in these two morphological characters only marginally affected pollinator efficiency. The strongest direct positive effect on pollen receipt in this study was pollinator visitation rate.
That corolla length and width had only weak effects on pollinator efficiency mirrors previous studies with I. aggregata. Campbell (1989a)
found only a weak relationship between pollen receipt by flowers of I. aggregata and corolla length and width. While floral shape (e.g., corolla length, corolla width) is thought to reflect, at least in part, selection through increased pollinator efficiency (e.g., Nilsson, 1988
; Schemske and Horvitz, 1989
), a surprising number of studies have not found a relationship between corolla shape and pollinator efficiency at depositing pollen on the stigmas of flowers (e.g., Murcia, 1990
; Young and Stanton, 1990
; Wilson, 1995
; Wilson and Thomson, 1996
). Corolla shape may reflect selection through male function (e.g., pollen removal, pollen export) more so than female function (Young and Stanton, 1990
; Campbell et al., 1991
; Connor et al., 1995
). We did not estimate the effect of corolla length and width on pollen export in this study. However, if pollen export is strongly affected by corolla shape, this effect may help explain the high levels of phenotypic variation in these characters in natural populations (Campbell, 1991
), especially if the strength of selection on floral morphology through male function varies from year to year.
One caveat about interpretation of our results is that path analysis is conditional upon which variables are included in the models, and the addition of other variables could change the magnitude and significance of the path coefficients in this study (Mitchell, 1993
). We included corolla length, corolla width, and plant height in the path models because these characters have been shown in previous studies to at least marginally affect pollinator visitation rate and/or pollinator efficiency in I. aggregata (Campbell, 1989a
; Wolf and Hainsworth, 1990
; Campbell et al., 1991
). However, other characters, such as stigma exsertion, nectar production rate, and flowering date, also affect pollinator visitation and/or pollen receipt (e.g., Campbell, 1989a
; Mitchell, 1992
), and we cannot rule out the importance of these characters in this system. Given these limitations, with our simple model we were still able to account for 24% of the variation in pollinator visitation rate and 36% of the variation in pollen receipt. More studies are needed to quantify the general relationship between pollinator visitation rate and pollen receipt across varying floral morphology and pollinator efficiency to test the generality of the positive link between pollinator visitation rate and pollen receipt.
|
FOOTNOTES |
|---|
5 Present address: Department of Ecology and Evolutionary Biology, University of Tennessee, Knoxville, Tennessee 37996 USA
6 Author for correspondence: Institute of Ecology, Ecology Building, University of Georgia, Athens, Georgia 30602 USA (Tel.: 706-583-8233; Fax: 706-542-4819; e-mail: rirwin{at}uga.edu
)
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Arnold R. M. 1982 Pollination, predation and seed set in Linaria vulgaris (Scrophulariaceae). American Midland Naturalist 107: 360-369[CrossRef][ISI]
Barrett S. C. H. 2002 Sexual interference of the floral kind. Heredity 88: 154-159[CrossRef][ISI][Medline]
Buchman S. L. G. P. Nabhan 1996 The forgotten pollinators. Island, Washington, D.C., USA
Cahill J. F. J. P. Castelli B. B. Casper 2001 The herbivory uncertainty principle: visiting plants can alter herbivory. Ecology 82: 307-312[ISI]
Cahill J. F. J. P. Castelli B. B. Casper 2002 Separate effects of human visitation and touch on plant growth and herbivory in an old-field community. American Journal of Botany 89: 1401-1409
Campbell D. R. 1989a Measurements of selection in a hermaphroditic plant: variation in male and female pollination success. Evolution 43: 318-334[CrossRef][ISI]
Campbell D. R. 1989b Inflorescence size: test of the male function hypothesis. American Journal of Botany 76: 730-738[CrossRef][ISI]
Campbell D. R. 1991 Effects of floral traits on sequential components of fitness in Ipomopsis aggregata. American Naturalist 137: 713-737[CrossRef][ISI]
Campbell D. R. 1996 Evolution of floral traits in a hermaphroditic plant: field measurements of heritabilities and genetic correlations. Evolution 50: 1442-1453[CrossRef][ISI]
Campbell D. R. K. J. Halama 1993 Resource and pollen limitations to lifetime seed production in a natural plant population. Ecology 74: 1043-1051[CrossRef][ISI]
Campbell D. R. N. M. Waser M. V. Price 1994 Indirect selection of stigma position in Ipomopsis aggregata via a genetically correlated trait. Evolution 48: 55-68[CrossRef][ISI]
Campbell D. R. N. M. Waser M. V. Price E. A. Lynch R. J. Mitchell 1991 Components of phenotypic selection: pollen export and flower corolla width in Ipomopsis aggregata. Evolution 45: 1458-1467[CrossRef][ISI]
Conner J. K. R. Davis S. Rush 1995 The effect of wild radish floral morphology on pollination efficiency by four taxa of pollinators. Oecologia 104: 234-245[CrossRef][ISI]
Costanza R. et al 1997 The value of the world's ecosystem services and natural capital. Nature 387: 253-260[CrossRef][Medline]
de Jong T. J. N. M. Waser M. V. Price R. M. Ring 1992 Plant size, geitonogamy and seed set in Ipomopsis aggregata. Oecologia 89: 310-315[ISI]
Fetscher A. E. J. R. Kohn 1999 Stigma behavior in Mimulus aurantiacus (Scrophulariaceae). American Journal of Botany 86: 1130-1135
Galen C. M. L. Stanton 1989 Bumble bee pollination and floral morphology: factors influencing pollen dispersal in the alpine sky pilot, Polemonium viscosum (Polemoniaceae). American Journal of Botany 76: 419-426[CrossRef][ISI]
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, 31–45. Van Nostrand Reinhold, New York, New York, USA
Greyson R. I. S. S. Tepfer 1967 Emasculation effects on stamen filament on Nigella hispanica and their partial reversal by gibberellic acid. American Journal of Botany 54: 971-976[CrossRef][ISI]
Hayduk L. A. 1987 Structural equation modeling with LISREL. Essentials and advances. The Johns Hopkins University Press, Baltimore, Maryland, USA
Inouye D. W. D. E. Gill M. R. Dudash C. B. Fenster 1994 A model and lexicon for pollen fate. American Journal of Botany 81: 1517-1530[CrossRef][ISI]
Jones K. N. J. S. Reithel 2001 Pollinator-meditated selection on a flower color polymorphism in experimental populations of Antirrhinum (Scrophulariaceae). American Journal of Botany 88: 447-454
Kearns C. A. D. W. Inouye 1993 Techniques for pollination biologists. University of Colorado Press, Niwot, Colorado, USA
Kingsolver J. G. D. W. Schemske 1991 Path analyses of selection. Trends in Ecology and Evolution 6: 276-280
Kiss H. G. R. E. Konig 1990 Emasculation effects on filament growth in Ipomea nil (Convulvulaceae). American Journal of Botany 77: 41-45
Li C. C. 1975 Path analysis: a primer. Boxwood, Pacific Grove, California, USA
Loehlin J. C. 1987 Latent variable models. Lawrence Erlbaum, Hillsdale, New Jersey, USA
Mitchell R. J. 1992 Testing evolutionary and ecological hypotheses using path analysis and structural equation modeling. Functional Ecology 6: 123-129
Mitchell R. J. 1993 Path analysis: pollination. In S. M. Scheiner and J. Gurevitch [eds.], Design and analysis of ecological experiments, 211–231. Chapman and Hall, New York, New York, USA
Mitchell R. J. N. M. Waser 1992 Adaptive significance of Ipomopsis aggregata nectar production: pollination success of single flowers. Ecology 73: 633-638[CrossRef][ISI]
Murcia C. 1990 Effect of floral morphology and temperature on pollen receipt and removal in Ipomoea trichocarpa. Ecology 71: 1098-1109[CrossRef][ISI]
Myers R. H. 1990 Classical and modern regression with applications, 2nd ed. PWS Kent, Boston, Massachusetts, USA
Nilsson L. A. 1988 The evolution of flowers with deep corolla tubes. Nature 334: 147-149[CrossRef]
Parker I. M. 1997 Pollinator limitation of Cytisus scoparius (scotch broom), an invasive exotic shrub. Ecology 78: 1457-1470[ISI]
Petraitis P. S. D. E. Dunham P. H. Niewiarowski 1996 Inferring multiple causality: the limitations of path analysis. Functional Ecology 10: 421-431[CrossRef][ISI]
Pleasants J. M. 1983 Nectar production patterns in Ipomopsis aggregata (Polemoniaceae). American Journal of Botany 70: 1468-1475[CrossRef][ISI]
Rademaker M. C. J. T. J. de Jong P. G. L. Klinkhamer 1997 Pollen dynamics of bumble-bee visitation on Echium vulgare. Functional Ecology 11: 554-563[CrossRef][ISI]
Schemske D. W. C. C. Horvitz 1989 Temporal variation in selection on a floral character. Evolution 43: 461-465[CrossRef][ISI]
Tepedino V. J. 1979 The importance of bees and other insect pollinators in maintaining floral species composition. Great Basin Naturalist Memoirs 3: 139-150
Waser N. M. 1978 Competition for hummingbird pollination and sequential flowering in two Colorado wildflowers. Ecology 59: 934-944[CrossRef][ISI]
Waser N. M. M. V. Price 1989 Optimal outcrossing in Ipomopsis aggregata: seed set and offspring fitness. Evolution 43: 1097-1109[CrossRef][ISI]
Waser N. M. M. V. Price 1991 Reproductive costs of self-pollination in Ipomopsis aggregata (Polemoniaceae): are ovules usurped?. American Journal of Botany 78: 1036-1043[CrossRef][ISI]
Wilson P. 1995 Selection for pollination success and the mechanical fit of Impatiens flowers around bumblebee bodies. Biological Journal of the Linnean Society 55: 355-383[CrossRef]
Wilson P. J. D. Thomson 1996 How do flowers diverge?. In D. G. Lloyd and S. C. H. Barrett [eds.], Floral biology, 88–111. Chapman and Hall, New York, New York, USA
Wolf L. L. F. R. Hainsworth 1990 Non-random foraging by hummingbirds: patterns of movement between Ipomopsis aggregata (Pursch) V. Grant inflorescences. Functional Ecology 4: 149-157
Wolf P. G. D. R. Campbell 1995 Hierarchical analysis of allozymic and morphometric variation in a montane herb, Ipomopsis aggregata (Polemoniaceae). Journal of Heredity 86: 386-394
Wright S. 1921 Correlation and causation. Journal of Agricultural Research 20: 557-585
Wright S. 1934 The method of path coefficients. Annals of Mathematics and Statistics 5: 161-215[CrossRef]
Young H. J. 1988 Differential importance of beetle species pollinating Dieffenbachia longispatha (Araceae). Ecology 69: 832-844[CrossRef][ISI]
Young H. J. M. L. Stanton 1990 Influences of floral variation on pollen removal and seed production in wild radish. Ecology 71: 536-547[CrossRef][ISI]
This article has been cited by other articles:
|
|
 |
|
  C. Alonso Pollination success across an elevation and sex ratio gradient in gynodioecious Daphne laureola Am. J. Botany, August 1, 2005; 92(8): 1264 - 1269. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
