Plant Ecology 172: 219–225, 2004. © 2004 Kluwer Academic Publishers. Printed in the Netherlands.

219

Lifetime fecundity and floral variation in Tuberaria guttata (Cistaceae), a Mediterranean annual Javier Herrera Departamento de Biología Vegetal y Ecología, Universidad de Sevilla. Apdo 1095, E-41080 Sevilla, Spain; (e-mail: [email protected]) Received 23 December 2002; accepted in revised form 26 June 2003

Key words: Allometry, Corolla, Mediterranean, Ovules, Petals, Stamens

Abstract In Tuberaria guttata, petal length, ovule number, and seeds per capsule raised steeply with increasing plant size 共respectively, in the ranges 6-11 mm, 40-100, and 20-80兲, while the number of stamens varied relatively little 共14-20兲. All flowers set fruit, and the rates of embryo abortion were independent of plant size and low on average. Individual fecundities had a markedly right-skewed frequency distribution 共in the ranges 1-20 capsules and 20-1500 seeds per plant兲, which issued not only from plant size and flower production being positively correlated, but also from per-flower ovule numbers being directly proportional to plant size. Correlated variation of plant and ovary sizes amplified among-plant inequalities regarding fecundity; allowed larger plants to set ca. 50% more seed than expected on the basis of flower number only; and caused the slope of the size-fecundity relationship to be considerably steeper 共at the population level兲 than if ovule number was a fixed trait. Corolla, ovary and androecium plasticity in Tuberaria are discussed in terms of environmental effects and developmental constraints.

Introduction In plant populations, individuals are often unequal as regards size and reproductive capacity. This has stimulated research on the effects of density and competition on plant reproductive performance 共Weiner 1988; Fone 1989; Kadmon and Shmida 1990; Rice 1990; Weiner and Thomas 1992兲. For perennials, reproductive inequalities may or not translate into varying ‘costs of reproduction’ 共e.g., reduced future growth or increased mortality; Horvitz and Schemske 1988; Reekie and Bazzaz 1992; Galen 1994; Primack and Stacey 1998兲, but for annuals seed output is always equivalent to lifetime fecundity and can be an acceptable surrogate of fitness. Studies of reproductive yield variation among conspecifics often emphasize the relationship between plant vegetative size and the number of flowers produced. In contrast, floral attributes have more often

been ignored as a source of individual differences in fecundity 共see however Burd, 1999兲. A likely reason for this neglect is that, because of form-function relationships which are often strict, flowers are plausible subjects for stabilizing selection to render them developmentally stable and less variable in form and size than, for example, leaves 共Conner and Sterling 1995; Sherry and Lord 1996; Armbruster et al. 1999兲. Furthermore, strong intrafloral correlations among organs can be expected to produce a single, integrated floral phenotype for each species 共Berg 1960; Stebbins 1974; Conner and Via 1993; Cresswell 2000; Herrera 2001; C.M. Herrera 2002; but see Wilson 1995兲. Therefore, reproductive inequalities among conspecifics are much more likely to arise from differences in flower number than from floral variability. Exceptional in this regard, however, could be those taxa in which the form-function relationship of flow-

220 ers is not strict and stamen or ovule numbers are not fixed genetically. It could be hypothesized that, in such species, the variable nature of flowers may contribute to individual differences in fecundity. In the present study I test this hypothesis with Tuberaria guttata, a selfing annual from Mediterranean scrublands. The issue of floral variability and its bearing on lifetime fecundity is addressed, and floral correlates of plant size investigated. The species 共once known as T. variabilis兲 was chosen because it has considerable variation as regards many phenotypic traits of flowers, inflorescences, and leaves.

dehisced capsules readily eject the minute 共0.5 mm long and 0.05 mg in mass兲, dust-like seeds. Fieldwork was carried out during March-April in an area of approximately one Ha near the town of Aznalcázar 共Andalucía, South Spain兲. The study population was on an insolated, flat area with fine sand at 40 m ASL where the dominant vegetation are extensive Pinus pinea woodlands mixed with scrub. Rainfall was close to average before and during the study, with precipitation totaling 463 mm from October to April 共the long-term mean for this period is 473 mm兲. Within-plant variations (position effects)

Material and methods Study species The genus Tuberaria 共Dunal兲 Spach 共known also as Xolantha Raf., Cistaceae兲 includes short-lived perennials and annuals with a Mediterranean distribution. Gallego 共1993兲 recognized nine species in the Iberian Peninsula, of which the annual T. guttata 共L.兲Fourr. 共Tuberaria, hereafter兲 is the most widely distributed. The plant is common in arid habitats ranging from evergreen-oak and pine woodlands through scrub clearings, grasslands and roadsides. In contrast with perennials, this taxon never shows the root-associated fungal structures 共ascocarps, i.e., subterranean Ascomycetes fruiting bodies兲 alluded to by its generic name. Seeds germinate from December to January and plants overwinter as rosettes of leaves 20-100 mm in diameter which, by late March, produce a straight inflorescence. Except in relatively large plants these racemes remain unbranched and open a single, yellow flower per day with the dish-bowl morphology typical in the Cistaceae. Flowers do not produce much pollen 共5500 grains per flower on average; Herrera 1992兲, secrete no nectar, and attract few insect visitors although, occasionally, pollen-collecting solitary bees, beetles, and Syrphid flies may be seen. The one-day flowers drop their petals at noon and then the stamens are pressed against the gynoecium by the closing sepals. Self-pollination often ensues since, in contrast with woody relatives in the Cistaceae 共e.g., Cistus; Talavera, Gibbs and Herrera 1993兲, Tuberaria is self-compatible. Flowering lasts in most populations until mid or late April, and capsule dehiscence occurs from April to May. When shaken,

Twenty medium-sized plants about to flower were individually tagged and used to investigate changes in per-flower stamen and ovule numbers during inflorescence development. On each of four dates covering the population flowering period 共March 20 and 31; April 8 and 15兲, the only flower to open that day on each plant was collected, placed in a numbered vial and taken to the laboratory. For simplicity, these will be referred in the Results by their position on the inflorescence axis 共e.g., first, mid-low, mid-high, and last兲, rather than by the date on which they were collected. Flowers were dissected and the numbers of stamens and ovules on each counted. Among-plant variations In order to keep within-plant variability controlled, and since the focus of this study was mainly at the variations of fecundity 共both absolute and relative兲 that occurred among plants, only the first flower that opened on each individual was sampled for counts and measurements 共i.e., the one at the bottom of the raceme兲. The allometric relationship that existed between individual plants and their flowers was investigated by noting the distance from ground level to the first flower 共height, hereafter兲 in mm, and the length of one randomly chosen petal in 42 individuals. Petals were placed between two glass slides and measured to the nearest 0.1 mm with digital calipers. Besides, a subsample of plants was taken whole 共roots included兲, then measured and weighed. Mass and height were tightly correlated 共r ⫽ 0.829, df ⫽ 18, p ⬍ 0.001兲. At the onset of flowering, the relationship between plant height and floral phenotypic gender was investigated. Twenty randomly chosen individuals were

221 measured, and the flower that had opened first collected to count stamens and ovules under a dissecting microscope. Later in the season, when capsules were ripe, another set of 20 randomly chosen plants was used to assess whether a relationship existed between plant size and seed abortion rates. After noting plant height, the number of seeds per fruit and the proportion of ovules that had eventually become seeds were determined for the most-basal fruit on each plant. Undeveloped ovules were easily differentiated from seeds on their smaller size and lighter color. The relationship between plant size and overall plant fecundity was studied when flowering had finished completely at the population. Ten 0.5 ⫻ 0.5 m plots were distributed at regular intervals across the site and surveyed for senescing Tuberaria. All plants detected 共N ⫽ 256兲 were measured, their fruits counted, and two different estimates of seed production per plant computed. The first one was simply the product of fruit number by the mean number of seeds per capsule as determined in a previous stage of the study 共see preceding paragraph兲. In so doing, per-flower ovule number was assumed a constant and variability regarding a major component of success as a female deliberately ignored. In contrast, the second estimate accounted for this variability by calculating seed output as the product of fruit number, ovule number, and the average rate of ovule transformation into seed. Ovules per flower were predicted for each of the 256 plants from the empirical regression Ovules ⫽ 5.2 ⫹ 0.53*Plant Height共r2 ⫽ 0.745, F ⫽ 52.7, N ⫽ 20, p ⬍ 0.001兲, whereas the rate at which ovules transformed into seed was virtually constant and known. The importance of floral variations for lifetime fecundity was then calibrated by comparing this estimate with the one that assumed a constant number of ovules per flower. Scale relationships between plant height and the other variables under study 共e.g., petal size兲 were investigated by reduced major axis 共RMA兲 regression. This method is preferred over ‘common’ 共least squares兲 regression whenever both variables are, as here, subject to natural variation and measurement error 共Niklas 1994兲. The slope of the RMA regression equals that of a least squares regression divided by the correlation coefficient.

Figure 1. Variations of floral phenotypic gender within Tuberaria inflorescences. Boxes delimit plus-minus one standard error around the mean 共horizontal line兲, and vertical lines encompass the nonoutlier range. Means sharing one lower case letter in a graph are not significantly different at p ⫽ 0.05 共Tukey’s multiple comparison test兲. Sample size is 20 plants.

Results A MANOVA aimed at detecting changes in the phenotypic gender of flowers according to their position within the inflorescence revealed varying effects on male and female expression. The model was a randomized block MANOVA in which the number of stamens and that of ovules were the response variables, and predictor ones were the position of the flower within the inflorescence 共i.e., first, mid-low, mid-high, and last兲 as well as plant identity 共blocking variable兲. Ovules per flower were highly dependent on position 共SS ⫽ 7057, df ⫽ 3, F ⫽ 27.7, p ⬍ 0.001兲, which in turn affected stamens only weakly 共SS ⫽ 77, df ⫽ 3, F ⫽ 3.2, p ⬍ 0.05兲. The magnitude and direction of position effects can be appreciated in Figure 1. Within an inflorescence, flowers from the lower- and top-half respectively had 70 and 50 ovules on average. In contrast, means for stamen number in early- and late-opening flowers were similar and statistically undistinguishable. Respectively, short and tall Tuberaria plants had petals near 6 mm and 10 mm long. Petal length and plant height correlated positively 共r ⫽ 0.645, df ⫽ 40, p ⬍ 0.001兲, and the scatterplot of data 共Figure 2兲 revealed a linear relationship with a slope significantly different from zero 共untransformed data: ␣rma ⫽ 0.51, with lower and higher 95 % confidence intervals re-

222

Figure 2. Half-log plot displaying the relationship between petal size and plant height in Tuberaria. Best-fit line is given for guidance.

spectively L1 ⫽ 0.39 and L2 ⫽ 0.64; plant height is log-transformed in Figure 2 for clarity兲. As shown in Figure 3, stamen, ovule, and seed number increased with plant height. The correlation was very tight for ovules 共r ⫽ 0.863, df ⫽ 18, p ⬍ 0.001兲, less so for stamens 共r ⫽ 0.47, df ⫽ 18, p ⬍ 0.05兲. Besides, ovules raised steeply with plant size 共␣rma ⫽ 0.61; L1 ⫽ 0.46, L2 ⫽ 0.76兲 whereas stamen number increased with a shallow 共although significantly different from zero兲 slope 共␣rma ⫽ 0.12; L1 ⫽ 0.07, L2 ⫽ 0.17兲. Seeds per fruit, on the other hand, also correlated positively with plant height 共Figure 3; r ⫽ 0.741, df ⫽ 18, p ⬍ 0.001兲. There were ca. 40 seeds per capsule on average 共mean ⫾ standard error x ⫽ 42 ⫾ 6; N ⫽ 20兲, even though this had considerable variance related to plant size 共for large and small plants, respectively 80 and 20 seeds兲. Seed-set, defined as the average proportion of ovules that developed into seed within a flower, was high 共x ⫽ 0.92 ⫾ 0.02, N ⫽ 12兲 and rates of embryo abortion accordingly low. Seed-set did not relate to plant height 共r ⫽ 0.27, df ⫽ 10, ns兲 and, actually, was less variable 共CV ⫽ 10%兲 than ovule 共24%兲 or stamen number 共21%兲. In the study population most individuals were shorter than 100 mm and had five or less fruits, whereas a few large plants might set up to 20 capsules 共Figure 4兲. Overall, fruit number correlated tightly with plant height 共r ⫽ 0.94, df ⫽ 254, p ⬍

Figure 3. Phenotypic gender and fecundity as affected by plant height. A: stamen and ovule numbers refer to the first flower to open on each of 20 plants. B: the number of seeds per capsule applies to the lowest 共early-formed兲 fruit.

0.001兲. The two size-fecundity functions used to calculate total plant seed output 共i.e., one that dismissed, and another that accounted for variations in the size of the ovary; see Methods兲 produced contrasting results 共Figure 4兲. Fecundity increased with plant size at a considerably faster rate 共i.e., the regression line was steeper兲 if floral variation in ovule number was accounted for. Although the discrepancy among the two functions was negligible if medium-sized plants were being considered 共some 300 seeds/plant兲, dismissing floral variability underestimated seed yield by ca. 50% in large plants.

Discussion In Tuberaria, the number of ovules inside the ovary was found to decrease towards the top of an inflorescence, from 70 in flowers that developped at the lowest half to approximately 50 for those near the top.

223

Figure 4. Fecundity as affected by size in Tuberaria. The number of fruits in a sample of 256 plants is represented by dots. Lines represent two possible relationships between estimated seed output and size: in one 共dashed line兲, seeds per plant is the product of fruit number and the average number of seeds per capsule; in the other 共solid line兲, seed output is the product of fruit number, ovules per flower, and seed-set rate. The difference in slope between these two lines summarizes the effect of ovarian variability on individual fecundity.

Given that overall individual seed production was calculated from basal flowers and fruits, absolute plant fecundities are likely to be overestimated 共by around 16%兲 in the present study. Nevertheless, the fecundity of a plant relative to others should not be affected by positional effects to any great exent, since such effects are likely to occur in all plants regardless of their size. Size and fecundity correlated positively in the study population and presented a similarly rightskewed distribution. As a result, there were very few truly fecund individuals and many relatively infertile ones, a common situation in plants 共Harper 1977; Scheiner 1987; Weiner 1988; C.M. Herrera 1991; Weiner and Thomas 1992兲. Furthermore, since flowers self-pollinated regularly 共i.e., there was no pollen limitation兲 and embryos seldom failed to develop, potential fecundity was equivalent to realized fecundity. This is also commonplace in autogamous annuals in which lifetime seed production closely tracks down flower number 共Lloyd 1980; Wiens et al. 1987兲. The study species was unusual, however, in that both the number of flowers and their quality were affected by plant size, which included changes in the diameter of the corolla and the size of the ovary 共as ovule number兲. Whether corolla size has any significant importance for reproduction in this largely self-pollinat-

ing species is debatable, but the importance of perflower ovule number for lifetime fertility is clear. Floral organ variability has often been reported within plants or inflorescences 共for example, Bawa and Webb 1983; Kang and Primack 1991; Svensson 1992; Diggle 1994; Ashman and Hitchens 2000; Mazer and Dawson 2001兲, but a consistent link between plant size and flower size represents an unexpected finding 共cf. Berg 1960兲. Admittedly, correlational data cannot reveal whether the bases of this variability are genetic, environmental or both, which should be found out growing plants from small and large mothers in a controlled environment. The observed continuum of correlated plant-flower sizes discovered in Tuberaria could result, for example, from varying levels of inbreeding depression in the population, although this would presuppose the existence of a mixed mating system and would require additional research to be confirmed. More parsimoniously, continuous variation of floral quantitative traits in the study plant can be interpreted as a consequence of environmental stress and/or competition acting on an extremely plastic herb. Hot, dry environments have been shown to induce significant floral variation in Raphanus sativus 共Mazer 1992兲, Clarkia tembloriensis 共Holtsford and Ellstrand 1992兲, Epilobium angustifolium 共Carroll, Pallardy and Galen 2001兲, and Datura wrightii 共Elle and Hare 2002兲. Moreover, plant defoliation results in a reduction of petal area in Brassica napus 共Cresswell, Hagen and Woolnough 2001兲. Under this view, Tuberaria plants might produce flowers with ovaries/ corollas as large and massive as allowed by their developmental condition. On the other hand, and since in the Cistaceae the growth of petals is postponed relative to ovary development 共Nandi 1998兲, correlated variation of corolla diameter and ovule number can be hypothesized to arise from intrafloral ontogenetic constraints. A greenhouse study by Mazer and Dawson 共2001兲 on Clarkia unguiculata has reported a trend similar to that found here to more ovules being contained in the ovary as plant size increases. In Clarkia however, the number of pollen grains was also intensified. Pollen production was not quantified in the present study, but the observation that stamen number 共and volume; personal observation兲 remained relatively unaltered in large and small Tuberaria would agree with models predicting greater relative allocation to the female function with increasing resource availability 共Charlesworth and Charlesworth 1981; Lovett-Doust and

224 Cavers 1982; Lloyd and Bawa 1984; Jong and Klinkhamer 1994兲.

Acknowledgements The author thanks M. Arista and C.M. Herrera for comments on an earlier version of the manuscript. This study was funded by Plan Andaluz de Investigación 共Junta de Andalucía兲, and by grant BOS20000328 of the Spanish Dirección General de Enseñanza Superior e Investigación Científica.

References Armbruster W.S., di Stilio V.S., Tuxill J.D., Flores T.C. and Velasquez Runk J.L. 1999. Covariance and decoupling of floral and vegetative traits in nine neotropical plants: a re-evaluation of Berg’s correlation-pleiades concept. American Journal of Botany 86: 39–55. Ashman T.L. and Hitchens M.S. 2000. Dissecting the causes of variation in intra-inflorescence allocation in a sexually polymorphic species, Fragaria virginiana 共Rosaceae兲. American Journal of Botany 87: 197–204. Bawa K.S. and Webb C.J. 1983. Floral variation and sexual differentiation in Muntingia calabura 共Elaeocarpaceae兲, a species with hermaphrodite flowers. Evolution 37: 1271–1282. Berg R.L. 1960. The ecological significance of correlation pleiades. Evolution 14: 171–180. Burd M. 1999. Flower number and floral components in ten angiosperm species: an examination of assumptions about tradeoffs in reproductive evolution. Biological Journal of the Linnean Society 68: 579–592. Carroll A.B., Pallardy S.G. and Galen C. 2001. Drought stress, plant water status, and floral trait expression in fireweed, Epilobium angustifolium 共Onagraceae兲. American Journal of Botany 88: 438–446. Charlesworth D. and Charlesworth B. 1981. Allocation of resources to male and female functions in hermaphrodites. Biological Journal of the Linnean Society 15: 57–74. Conner J.K. and Via S. 1993. Patterns of phenotypic and genetic correlations among morphological and life-history traits in Wild Radish, Raphanus raphanistrum. Evolution 47: 704–711. Conner J.K. and Sterling A. 1995. Testing hypotheses of functional relationships: a comparative survey of correlation patterns among floral traits in five insect-pollinated plants. American Journal of Botany 82: 1399–1406. Cresswell J.E. 2000. Manipulation of female architecture in flowers reveals a narrow optimum for pollen deposition. Ecology 81: 3244–3249. Cresswell J.E., Hagen C. and Woolnough J.M. 2001. Attributes of individual flowers of Brassica napus L. are affected by defoliation but not by intraspecific competition. Annals of Botany 88: 111–117. Diggle P.K. 1994. The expression of andromonoecy in Solanum hirtum 共Solanaceae兲: phenotypic plasticity and ontogenetic contingency. American Journal of Botany 81: 1354–1365.

Elle E. and Hare J.D. 2002. Environmentally induced variation in floral traits affects the mating system in Datura wrightii. Functional Ecology 16: 79–88. Fone A.L. 1989. A comparative demographic study of annual and perennial Hypochaeris 共Asteraceae兲. Journal of Ecology. 77: 495–508. Galen C. 1994. Cost of reproduction in Polemonium viscosum – Phenotypic and genetic approaches. Evolution 47: 1073–1079. Gallego M.J. 1993. 3. Xolantha Raf. Pp 351-365 In: Castroviejo S., Aedo C., Cirujano S., Lainz M., Montserrat P., Morales R., Muñoz Garmendia F., Navbarro C., Paiva J. and Soriano C. 共eds兲, Flora Iberica Vol III. Real Jardín Botánico, Madrid, Spain. Harper J.L. 1977. Population biology of plants. Academic Press, London, UK. Herrera C.M. 1991. Dissecting factors responsible for individual variation in plant fecundity. Ecology 72: 1436–1448. Herrera C.M., Cerda X., Garcia M.B., Guitian J., Medrano M., Rey P.J. and Sanchez-Lafuente A.M. 2002. Floral integration, phenotypic covariance structure and pollinator variation in bumblebeepollinated Helleborus foetidus. Journal of Evolutionary Biology 15: 108–121. Herrera J. 1992. Flower variation and breeding systems in the Cistaceae. Plant Systematics and Evolution 179: 245–256. Herrera J. 2001. The variability of organs differentially involved in pollination, and correlations of traits in genisteae 共Leguminosae : Papilionoideae兲. Annals of Botany 88: 1027–1037. Holtsford T.P. and Ellstrand N.C. 1992. Genetic and environmental variation in floral traits afffecting outcrossing rate in Clarkia tembloriensis 共Onagraceae兲. Evolution 46: 216–225. Horvitz C.C. and Schemske D.W. 1988. Demographic cost of reproduction in a neotropical herb: an experimental field study. Ecology 69: 1741–1745. Jong T.J.D. and Klinkhamer P.G.L. 1994. Plant size and reproductive success through female and male function. Journal of Ecology. 82: 399–402. Kadmon R. and Shmida A. 1990. Patterns and causes of spatial variation in the reproductive success of a desert annual. Oecologia 83: 139–144. Kang H. and Primack R.B. 1991. Temporal variation of flower and fruit size in relation to seed yield in Celandine Poppy 共Chelidonium majus, Papaveraceae兲. American Journal of Botany 78: 711–722. Lloyd D.G. 1980. Sexual strategies in plants. I. An hypothesis of serial adjustment of maternal investment during one reproductive session. New Phytologist 86: 69–79. Lloyd D.G. and Bawa K.S. 1984. Modification of the gender of seed plants in varying conditions. Evolutionary Biology 17: 255–339. Lovett-Doust J. and Cavers P.B. 1982. Biomass allocation in hermaphrodite flowers. Canadian Journal of Botany 60: 2530–2534. Mazer S.J. 1992. Environmental and genetic sources of variation in floral traits and phenotypic gender in Wild Radish: consequences for natural selection. Pp 281- 325 In: Wyatt R. 共ed.兲, Ecology and evolution of plant reproduction. Chapman and Hall, New York, New York, USA. Mazer S.J. and Dawson K.A. 2001. Size-dependent sex allocation within flowers of the annual herb Clarkia unguiculata 共Onagraceae兲: ontogenetic and among-plant variation. American Journal of Botany 88: 819–831. Nandi O.I. 1998. Floral development and systematics of Cistaceae. Plant Systematics and Evolution 212: 107–134.

225 Niklas K.J. 1994. Plant allometry. University of Chicago Press, Chicago, Illinois, USA. Primack R. and Stacy E. 1998. Cost of reproduction in the pink lady’s slipper orchid, 共Cypripedium acaule, Orchidaceae兲: an eleven-year experimental study of three populations. American Journal of Botany 85: 1672–1679. Reekie E.G. and Bazzaz F.A. 1992. Cost of reproduction as reduced growth in genotypes of 2 congeneric species with contrasting life histories. Oecologia 90: 21–26. Rice K.J. 1990. Reproductive hierarchies in Erodium: effects of variation in plant density and rainfall distribution. Ecology 71: 1316–1322. Scheiner S.M. 1987. Size and fecundity hierarchies in an herbaceous perennial. Oecologia 74: 128–132. Sherry R.A. and Lord E.M. 1996. Developmental stability in flowers of Clarkia tembloriensis 共Onagraceae兲. Journal of Evolutionary Biology 9: 911–930. Stebbins G.L. 1974. Flowering plants: evolution above the species level. Belknap Press, Cambridge, Massachusetts, USA. Svensson L. 1992. Estimates of hierarchical variation in flower morphology in natural populations of Scleranthus annuus

共Caryophyllaceae兲, an inbreeding annual. Plant Systematics and Evolution 180: 157–180. Talavera S., Gibbs P.E. and Herrera J. 1993. The reproductive biology of Cistus ladanifer L. 共Cistaceae兲. Plant Systematics and Evolution 186: 123–134. Weiner J. 1988. The influence of competition on plant reproduction. Pp. 228-245 In: Lovett-Doust and Lovett-Doust 共eds兲, Plant reproductive ecology: patterns and strategies. Oxford Univ. Press, New York, New York, USA. Weiner J. and Thomas S.C. 1992. Competition and allometry in 3 species of annual plants. Ecology 73: 648–656. Wiens D., Calvin C.L., Wilson C.A., Davern C.I., Frank D. and Seavey S.R. 1987. Reproductive success, spontaneous embryo abortion, and genetic load in flowering plants. Oecologia 71: 501–509. 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.