| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
0 Laboratorio de Botánica, Facultad de Farmacia, Departamento de Biología Vegetal, Universidad de Santiago de Compostela, Campus Sur s/n, Santiago, E-15706 Spain
ABSTRACT
We investigated patterns of fruit and seed production on inflorescences of a population of Pancratium maritimum in northwest Spain over a 2-yr period. Initial findings showed that the earliest opening flowers on an inflorescence are more likely to set fruit and produce more seeds than later opening flowers and that this pattern is maintained throughout the flowering season. Supplementary pollination and flower-removal experiments were performed to investigate whether the observed pattern is attributable (a) to variation in pollen receipt, (b) to sequestration of resources by the earliest flowers on an inflorescence, and/or (c) to "architectural" limitations on the fruit/seed production of later flowers. Supplementary pollination did not improve fruit or seed production by late flowers in either of the 2 yr of study. In flower-removal experiments, the remaining flowers showed improved fruit set and mean number of seeds per flower, by comparison with flowers in the same position on control inflorescences. When all flowers except the latest third were removed, these showed fruit set and seed production similar to those of early flowers on control inflorescences. These results strongly suggest that the observed within-inflorescence patterns of fruit and seed production in P. maritimum are mainly attributable to competition for resources (i.e., explanation b), though other adaptive explanations cannot be ruled out.
Key Words: Amaryllidaceae • architectural effects • floral variation • fruit and seed production • Pancratium maritimum; • pollen limitation • resource allocation
In plants in which the flowers are grouped in inflorescences, consistent patterns of within-inflorescence variation in morphological and functional characteristics of flowers are commonly observed. Numerous studies have investigated relationships between reproductive success and within-inflorescence spatial position (proximal or distal) and/or time of opening (early or late) (Stephenson, 1980, 1981
; Wyatt, 1982
; Bawa and Webb, 1984
; Holtsford, 1985
; McKone, 1985
; Lee, 1988
; Solomon, 1988
; Schoen and Dubuc, 1990
; Berry and Calvo, 1991
; Herrera, 1991
; Ehrlén, 1992
; Karoly, 1992
; Goldingay and Whelan, 1993
; Obeso, 1993
; Vaughton, 1993
; Guitián, 1994
; Brunet, 1996
; Emms, 1996
; Guitián and Navarro, 1996
; Navarro, 1996
; Méndez, 1997
; Vaughton and Ramsey, 1997
; for a review, see Diggle, 1995
). In species with acropetal inflorescence development, the probability of fruit set and/or number of seeds per flower is often lower for distal/late-opening flowers than for proximal/early-opening flowers (Solomon, 1988
; Herrera, 1991
; Ehrlén, 1992, 1993
; Karoly, 1992
; Guitián, 1994
; Guitián and Navarro, 1996
; Navarro, 1996
). Less commonly, fruit set is highest in intermediate flowers (Sutherland, 1987
) or in distal/late-opening flowers (Goldingay and Whelan, 1993
). Some studies of species with basipetal inflorescence development have found higher fruit and seed set in distal/early-opening flowers (Brunet, 1996
), while others have found higher fruit and seed set in intermediate flowers (Berry and Calvo, 1991
) or proximal/late-opening flowers (Vaughton, 1993
; Vaughton and Ramsey, 1997
). Rarely, there have been reports of species in which fruit and seed production are independent of position within the inflorescence (Zimmerman and Aide, 1989
).
Various nonexclusive hypotheses have been put forward to explain such patterns of within-inflorescence variation in female reproductive success.
1) The "resource competition hypothesis" (Stephenson, 1981
; Bawa and Webb, 1984
; Nakamura, 1986
; Lee, 1988
; Thomson, 1989
; Guitián, 1994
) postulates that the observed patterns are attributable to competition among the ovaries of an inflorescence for a limited amount of resources; fruits initiated early, and/or located close to the source of nutrients, sequester more resources than fruits initiated later and/or located more distally (Lee, 1988
).
2) The "non-uniform pollination hypothesis" (Thomson, 1985, 1989
; Lee, 1988
; Berry and Calvo, 1991
; Goldingay and Whelan, 1993
; Brunet and Charlesworth, 1995
) postulates that the observed patterns may be attributable to variation in pollen receipt over the inflorescence flowering period. Specifically, the relatively low fruit or seed sets of certain flowers on the inflorescence may be attributable to insufficient quantity or quality of pollen (Lee, 1988
).
3) The "architectural effects hypothesis" (Wyatt, 1982
; Lee, 1988
; Thomson, 1989
; Wolfe, 1992
; Diggle, 1992, 1995, 1997
) postulates that the observed patterns are attributable to intrinsic ("architectural") limitations on the reproductive success of flowers in different positions within the inflorescence (Diggle, 1995
, and references therein). This hypothesis is based on studies in which the fruit set of late-opening flowers is not affected by the removal of early-opening flowers or the prevention of their fruiting (Sutherland, 1987
; Berry and Calvo, 1991
).
In the present study we investigated patterns of fruit set and seed production among flowers within umbels of Pancratium maritimum L. (Amaryllidaceae) over two reproductive seasons. Our aim was to investigate the extent to which these three hypotheses explain the patterns observed in an umbel-type inflorescence, within which there is temporal but not spatial variation in opening time among flowers (i.e., flowers open at different times but do not vary appreciably as regards distance from the growth axis). Species with inflorescences of this type have been little studied to date (though see Wyatt, 1980
). Because a previous study revealed that fruit set varies over the flowering season (Medrano, Guitián, and Guitián, 1999
), we took account of possible effects of timing in the population flowering period. Specifically, we aimed to determine whether fruit set and seed set are affected by position within the flowering order on the umbel, and if so (1) whether the observed patterns are dependent on timing in the flowering period and (2) to what extent pollination, resource availability, and/or architectural effects influence the female reproductive success of flowers at different positions within the inflorescence.
MATERIALS AND METHODS
The plant and the study area
Pancratium maritimum L. is a bulbous perennial geophyte that occurs on dunes and other sandy coastal habitats. In Europe and North Africa it is present in Morocco, the Iberian Peninsula, and France. In the study area (see below), the flowering period is June to September. Pancratium maritimum has an umbel-type inflorescence with 2–14 white flowers, which develop centripetally on a scape ~15 cm long. Plants generally produce one inflorescence per season (occasionally two). Anthesis occurs at dusk and the flower remains open for ~36 h. Depending on the number of flowers on the inflorescence, the time elapsed between anthesis of the first and last flower ranges from 3 to 15 d (since flowers open sequentially, generally one per day, though sometimes two or rarely three per day; M. Medrano, unpublished data). In a previous study, P. maritimum was found to be self-compatible and to produce fruits by autonomous autogamy; furthermore, fruit set (though not seed set) varied over the flowering season (Medrano, Guitián, and Guitián, 1999
). The mean number of ovules per flower is 56.6 (N = 74 flowers, SD = 9.7). The fruit is a capsule with loculicid dehiscence, containing on average 13.6 seeds (N = 341, SD = 9.8).
The study was carried out in 1996 and 1997 in a population of P. maritimum in the Corrubedo Dune Complex National Park in the southwest of A Coruña Province in Galicia (northwest Spain). At this site, P. maritimum occurs on stabilized sand as part of chamaephyte communities dominated by species including Helichrysum picardii var. virescens, Artemisia crithmifolia, Scophularia frutescens, and Iberis procumbens.
At the nearby weather station of Santa Eugenia de Riveira (25 m a.s.l.), mean temperature is 14.6°C, and mean annual rainfall 1244 mm (Carballeira et al., 1983
).
Experimental design
To characterize patterns of fruit set in the study population, and to investigate whether these patterns vary over the flowering period (12 and 18 wk in 1996 and 1997, respectively), we divided the flowering period into subperiods, as shown in Fig. 1. Note that the subperiods defined in each year do not correspond, since flowering patterns differed markedly between the 2 yr of study. Note also that fruit set in subperiod D of 1997 was zero in all experimental groups and controls, and this subperiod was therefore excluded from the analyses. Patterns of seed set within inflorescences were investigated in the same plants as those used for the study of fruit set, but only in subperiods 1996-A, 1996-B, 1997-E, and 1997-F. At the end of the reproductive season in both years of study, the study population suffered intense herbivory by larvae of the moth Brithys crini subsp. pancratii (Noctuidae), with the result that sample sizes in 1996-C and 1997-G were insufficient for analysis of patterns of seed set.
|
Open pollination: seed set
We collected a total of 33 "complete" infructescences (i.e., infructescences that had not lost any full-size fruits as a result of herbivory during the maturation process) from the naturally pollinated nonmanipulated plants monitored in 1996 (six infructescences from subperiod A, 27 from subperiod B; N = 139 fruits), and a total of 60 complete infructescences from the naturally pollinated nonmanipulated plants monitored in 1997 (36 from subperiod E, 24 from subperiod F; N = 247 fruits). In all cases infructescences were collected 60–80 d after anthesis of the first flower. The infructescences were taken to the laboratory, and the number of mature and aborted seeds in each fruit was counted with the aid of a stereomicroscope.
Nonuniform pollination hypothesis: fruit set
If within-inflorescence patterns in female reproductive success are attributable to differences in amount of pollen received, supply of pollen to all flowers should lead to uniform female reproductive success. To test for this possibility, we carried out supplementary pollination (SP) experiments in 1996 and 1997, randomly selecting 14–19 plants in flower in each subperiod (one inflorescence per plant; 1996-A, 14 plants; 1996-B, 14 plants; 1996-C, 15 plants; 1997-E, 19 plants; 1997-F, 19 plants; 1997-G, 18 plants). As each flower opened, we numbered and pollinated it with a brush containing abundant xenogamous pollen, obtained by collecting recently opened stamens from 10 to 15 plants located at least 5 m away from the recipient plant. The brush-pollination was repeated 24 h later. The flowers were subsequently monitored weekly, and fruit set was noted 20–30 d after anthesis.
Nonuniform pollination hypothesis: seed set
We collected a total of 16 complete infructescences (i.e., infructescences not affected by herbivory) from the supplementary-pollinated plants monitored in 1996 (eight infructescences from subperiod A, eight from subperiod B; N = 59 fruits), and a total of 30 complete infructescences from the supplementary-pollinated plants monitored in 1997 (16 from subperiod E, 14 from subperiod F; N = 185 fruits). The infructescences were taken to the laboratory, and the number of mature and aborted seeds in each fruit was counted with the aid of a stereomicroscope.
Architectural effects vs. resource competition: fruit set
To discriminate between architectural effects and resource competition among flowers within inflorescences, flower-removal experiments were performed in 1996 and 1997. The 1996 experiments were designed to identify possible variation in these effects over the flowering season: from each inflorescence (one inflorescence per plant; 14 plants in 1996-A, 16 plants in 1996-B, 16 plants in 1996-C), we removed the first three flowers to open (treatment FR3e). The 1997 experiments were carried out only in subperiod E: we randomly selected 34 inflorescences (on 34 plants) and removed the first two-thirds of the flowers to open (16 inflorescences; treatment FRe+i) or the last two-thirds of the flowers to open (18 inflorescences; treatment FRi+l). In all cases (i.e., both in 1996 and 1997), the flowers removed were cut within 24–72 h of anthesis, and the remaining flowers in the inflorescence were brush-pollinated (as described above for treatment SP), to allow us simultaneously to rule out the possibility of effects due to deficient pollination. In both years, the controls were inflorescences from the same subperiod from the SP experiments. [A detailed explanation of the basis for experimental designs of this type (FR3e, FRe+i, FRi+l, controls) is given by Diggle (1997)]. The flowers were monitored weekly, and whether each set fruit was noted 20–30 d after anthesis.
Architectural effects vs. resource competition: seed set
We collected a total of 23 complete infructescences from the plants subjected to treatment FR3e in 1996 (11 infructescences from subperiod A, 12 from subperiod B; N = 65 fruits). In 1997, we collected 16 complete infructescences from the inflorescences subjected to treatment FRe+i, and 16 from the inflorescences subjected to treatment FRi+l (N = 78 fruits). The infructescences were taken to the laboratory, and the number of mature and aborted seeds in each fruit was counted with the aid of a stereomicroscope.
Data analysis
In all analyses of factors influencing fruit/seed set, only inflorescences that produced at least one fruit were considered. To allow consideration of inflorescences with different numbers of flowers, flower "position" was in all cases defined as "early" (the earliest third to open), "intermediate" (the middle third), or "late" (the last third). For example, flower number 3 in an 11-flower inflorescence was classified as "early," and flower number 4 as "intermediate" (see Berry and Calvo, 1991
). Unless otherwise stated, statistical significance is taken to be indicated by P values of <5%.
Fruit set
To investigate whether the probability that a flower will set fruit varies with position in the inflorescence and, if so, whether the effect of position varies over the flowering season, we used maximum-likelihood logistic regression (CATMOD; SAS, 1992
), with the independent variables "position" and "position x subperiod." The data for the 2 yr and for the two modes of pollination (natural or supplementary) were analyzed separately. Subsequent pairwise comparisons between positions were done with G tests; since these were multiple comparisons, the significance level was corrected for the number of tests (Bonferroni method; Sokal and Rohlf, 1995
). G tests were also used to compare fruit set between flowers from inflorescences subjected to flower removal and their corresponding controls.
Seed set
For each position (early, intermediate, and late) and each inflorescence of each treatment (control, supplementary pollination, flower-removal treatments), we calculated mean number of seeds per flower and mean seed-to-ovule ratio, taking flowers that did not set seed into account. For between-position comparisons of mean number of seeds per flower and mean seed-to-ovule ratio, we used nonparametric tests in all cases, since the data were markedly non-normal even after standard transformations (largely because of frequent zero seed set among "late" flowers). Specifically, the effects of position and pollination type on mean number of seeds per flower and mean seed-to-ovule ratio were investigated by Friedman's analysis of variance, with subsequent pairwise comparisons between positions by Tukey tests for nonparametric randomized-block analysis of variance (Zar, 1996
). The effects of flower removal on seed production by the remaining flowers were investigated with Kruskal-Wallis tests.
RESULTS
Open pollination
Fruit set
Data on the fruit set of naturally pollinated flowers (considering only those inflorescences that produced at least one fruit) are summarized for each position and each subperiod in Figs. 2A and 3A. The results of maximum-likelihood logistic regression indicate that in both years the factor "position" had significant effects on the probability that a flower set fruit (1996: 
2 = 87.1, df = 2, P < 0.0001; 1997: 2 = 134.3, df = 2, P < 0.0001); the interaction "position x subperiod" had a significant effect in 1997 (2 = 24.9, df = 4, P = 0.0001) but not in 1996 (2 = 7.3, df = 4, P = 0.12). The results of G tests indicate that the fruit set of early flowers was significantly higher than that of intermediate and late flowers, except in 1997 subperiods E and F, when fruit set did not differ significantly between early and intermediate flowers (Figs. 2A, 3A). In both years and all subperiods, the fruit set of intermediate flowers was significantly higher than that of late flowers (Figs. 2A, 3A).
|
|
2 = 7.7, P = 0.021; 1996-B: 2 = 10.9, P = 0.004; 1997-E: 2 = 7.6, P = 0.022; 1997-F: 2 = 21.3, P < 0.0001). Similarly, mean seed-to-ovule ratio varied significantly with position in all subperiods except 1997-E (1996-A: 2 = 7.7, P = 0.021; 1996-B: 2 = 9.2, P = 0.01; 1997-E: 2 = 4.5, P = 0.108; 1997-F: 2 = 19.8, P = 0.0001). Pairwise comparisons with Tukey tests indicated that neither mean number of seeds per flower nor mean seed-to-ovule ratio differed significantly either between early and intermediate flowers or between intermediate and late flowers, in any of the subperiods considered (with the exception of 1997 subperiod F, in which both variables were significantly higher for intermediate flowers than for late flowers) (Fig. 5A). However, both variables (mean number of seeds per flower and mean seed-to-ovule ratio) were significantly higher for early flowers than for late flowers in subperiods A, B, and F (Figs. 4A, 5A).
|
|
2 = 44.9, df = 2, P < 0.0001; 1997: 2 = 87.1, df = 2, P < 0.0001). The interaction "position x subperiod" had a significant effect in 1997 (2 = 13.1, df = 4, P = 0.011) but not in 1996 (2 = 3.9, df = 4, P = 0.426). The results of G tests indicate that the fruit set of early and intermediate flowers was significantly higher than that of late flowers in all subperiods of both years, even with supplemental pollination, except in 1996 subperiod C, when fruit set did not differ significantly between intermediate and late flowers (Figs. 2B, 3B). However, the differences between early and intermediate flowers were not always significant (Figs. 2B, 3B).
Seed set
The among-position differences observed in seed set under open pollination were maintained when supplemental pollen was applied (Figs. 4B, 5B). Position had a significant effect on mean number of seeds per flower in all subperiods in both years, except in 1996 subperiod A (1996-A: 2 = 5.8, P = 0.055; 1996-B: 2 = 10.8, P = 0.005; 1997-E: 2 = 11.5, P = 0.003; 1997-F: 2 = 19, P = 0.0001). Similarly, position had a significant effect on mean seed-to-ovule ratio in all subperiods in both years (1996-A: 2 = 6.3, P = 0.043; 1996-B: 2 = 10.8, P = 0.005; 1997-E: 2 = 11.7, P = 0.003; 1997-F: 2 = 19.9, P < 0.0001). Pairwise comparisons with Tukey tests indicated that neither mean number of seeds per flower nor mean seed-to-ovule ratio differed significantly between early and intermediate flowers in any of the subperiods considered (Figs. 4, 5). Considering intermediate and late flowers, no significant differences in either mean number of seeds per flower or mean seed-to-ovule ratio were observed in 1996, except in subperiod A, in which the mean seed-to-ovule ratio was significantly higher in intermediate than in late flowers; in all subperiods of 1997, however, both variables were significantly higher for intermediate than for late flowers (Figs. 4B, 5B). Early flowers consistently showed significantly higher mean number of seeds per flower and mean seed-to-ovule ratio than late flowers, except in 1996-A, in which the P value obtained was slightly greater than 5% (Fig. 4B).
Architectural effects vs. resource competition
Fruit set
In the 1996 experiments, the three earliest flowers were removed (just after anthesis) from treated inflorescences. The mean number of flowers remaining on the inflorescence was 4.9 (range 2–7). In 1996 subperiods B and C, the removal of the three earliest flowers significantly improved the fruit set of both the intermediate and late flowers of those inflorescences (with respect to flowers in the same positions in control inflorescences) (Table 1); in 1996 subperiod A, however, no significant effect was observed (Table 1). In none of the three subperiods did the fruit set of intermediate flowers in treated inflorescences differ significantly from that of early flowers in control inflorescences (subperiod A: G = 0.29, P = 0.44; subperiod B: G = 5.27, P = 0.07; subperiod C: G = 0.85, P = 0.28).
|
|
|
|
Fruit and seed set under open pollination
Under open pollination the earliest opening flowers of inflorescences of Pancratium maritimum have a higher probability of setting fruit than later opening flowers. We detected clear differences between both "early" and "intermediate" flowers and "intermediate" and "late" flowers. Mean number of seeds per flower and mean seed-to-ovule ratio were also higher for early- than for late-opening flowers; however, statistically significant differences were only detected between "early" and "late" flowers. Moreover, these patterns were maintained in all subperiods of both years of study, although fruit set varied significantly among subperiods in 1996, in accordance with results reported previously (Medrano, Guitián, and Guitián, 1999
). In Petrocoptis grandiflora, the greater fruit set of early flowers is also maintained throughout the flowering season (Guitián and Navarro, 1996
).
Our findings are in agreement with previous studies of other hermaphroditic species with flowers grouped in inflorescences, including for example, Lupinus luteus (Van Stevenick, 1957
), Phaseolus vulgaris (Tamas et al., 1979
), Catalpa speciosa (Stephenson, 1979, 1980
), Asclepias tuberosa (Wyatt, 1980
), Yucca whipplei (Udovic and Aker, 1981
), Caesalpinia eriostachys (Bawa and Webb, 1984
), Calochortus leichtlinii (Holtsford, 1985
), Solanum carolinense (Solomon, 1988
), Lavandula stoechas (Herrera, 1991
), Lathyrus vernus (Ehrlén, 1992
), Banksia spinulosa (Vaughton, 1993
), Prunus mahaleb (Guitián, 1994
), Petrocoptis grandiflora (Guitián and Navarro, 1996
), and Aquilegia caerulea (Brunet, 1996
). Possible proximal causes of the decreasing female reproductive success observed within inflorescences of P. maritimum, together with possible ultimate functions of late flowers, are discussed below.
Nonuniform pollination hypothesis
Variation in the amount of pollen received does not explain the observed differences in fruit and seed set between early and late flowers, because the differences are maintained when abundant pollen is supplied to all flowers. The patterns observed under natural pollination were maintained under supplemental pollination: the probability of a flower setting fruit was higher for early and intermediate flowers than for late flowers, and both mean number of seeds per flower and mean seed-to-ovule ratio were higher for early and intermediate flowers. Similar results have been obtained with many other species (see, e.g., Holtsford, 1985
; Devlin, 1989
; Karoly, 1992
; Brunet, 1996
). In a study of Agave mckelveyana (a species in which fruit set under natural conditions is lower for the earliest and most basal flowers in the inflorescence), Sutherland (1987)
; found that most basal flowers aborted regardless of whether or not supplemental pollen was supplied. In Lathyrus vernus, Ehrlén (1992)
found that even the addition of outcross pollen had no beneficial effect on the fruit set of late flowers. In some species, by contrast, within-inflorescence patterns of fruit and seed production have been shown to be due to variation in the amount of pollen received (Berry and Calvo, 1991
; Goldingay and Whelan, 1993
).
Variations in fruit and seed set among flowers within inflorescences may be due to variation in the quality (i.e., source) of pollen arriving at flowers in the various positions and to subsequent selective abortion of fruits (for reviews see Stephenson, 1981
; Lee, 1988
). Variations in the quality of pollen received by flowers at different positions on the inflorescence have been related to directional movement of pollinators, to the order of development of flowers, and to the degree of within-inflorescence dichogamy (temporal overlap of male and female function) (Wyatt, 1982
; Nilsson, 1983
; Berry and Calvo, 1991
; Brunet and Charlesworth, 1995
; Brunet, 1996
). In species with protandrous flowers on racemose acropetal inflorescences, for example, it has been found that early flowers have a higher probability of receiving xenogamous pollen than later and more apical flowers because of the pollinators' tendency to move up the inflorescence (Wyatt, 1982
, and references therein). The present study did not include experiments designed to assess whether the observed pattern of decline in fruit and seed production in P. maritimum might be due to variation in pollen quality. However, two observations argue against this possibility. First, within inflorescences of P. maritimum there is virtually no overlap in flowering periods (i.e., most of the time there is only one flower open, though sometimes two or very rarely three flowers may open on the same day), so that the probability of geitonogamous pollen transfer is low. Second, previous studies have not detected significant differences between selfed and outcrossed flowers in either fruit and seed set (Medrano, Guitián, and Guitián, 1999
) or seed dry mass (M. Medrano, unpublished data); Pancratium maritimum therefore does not appear to show inbreeding depression (at least as regards fertilization, and development and maturation of fruits and seeds).
Architectural effects vs. resource competition
The results of our 1996 and 1997 flower-removal experiments indicate that removal of flowers from inflorescences of P. maritimum improves the fruit and seed set of the remaining flowers. This suggests that the flowers in an inflorescence compete for resources and that late flowers may act as "ovary reserves." Similar findings have been reported for many other species with flowers grouped in inflorescences, including Catalpa speciosa (Stephenson, 1980
), Calochortus leichtlinii (Holtsford, 1985
), Lavandula stoechas (Muñoz and Devesa, 1987
), Lathyrus vernus (Ehrlén, 1992
), Banksia spinulosa (Vaughton, 1993
), Prunus mahaleb (Guitián, 1994
), Aquilegia caerulea (Brunet, 1996
), and Petrocoptis grandiflora (Guitián and Navarro, 1996
).
In 1996, removing early flowers did not significantly increase fruit set, seed production, or seed-to-ovule ratio of either intermediate or late flowers, except for subperiod B. This may simply reflect the fact that infructescence sample sizes were small in this year. By contrast, the 1997 flower-removal experiments, with a larger sample size, showed significant effects.
If removal of "early" (or "early and intermediate") flowers increases the fruit set of "intermediate and late" (or "late" flowers) by only a limited amount (i.e., to a level below that of control flowers), one possible explanation is that an "architectural" effect is acting, i.e., that later opening flowers have intrinsically lower maximum fruit set than early-opening flowers, as has been found in a number of species (Sutherland, 1987
; Goldingay and Whelan, 1993
), though it has also been suggested that prefertilization among-ovary inequalities in resource distribution, only reversible at early stages in inflorescence development, may be responsible (Solomon, 1988
). It has been suggested that architectural effects are attributable to reduced vascular supply to more distal positions, considered an inevitable consequence of indeterminate growth of the floral axis (Wolfe, 1992
; Diggle, 1995, 1997
). In P. maritimum, the latest opening flowers of the umbel are located in the same position with respect to the growth axis as the earliest opening flowers, but floral peduncle length, perianth size, and number of ovules are all greater in early flowers than in late flowers (M. Medrano, unpublished data). However, our present findings indicate that removal of early flowers increased the fruit set of later flowers to levels similar to those of early flowers of control inflorescences, arguing against the architectural effect hypothesis for P. maritimum. Similar results have been obtained for species including Catalpa speciosa (Stephenson, 1980
) and Banksia spinulosa (Vaughton, 1993
).
Specifically, we found that fruit production by intermediate flowers following elimination of the first three flowers in the inflorescence did not differ significantly from that of early flowers of control inflorescences. The results of our 1997 flower-removal experiments indicate that fruit and seed production after removal of two-thirds of the flowers in the inflorescence were independent of whether the flowers remaining were the earliest third or the latest third. Furthermore, when all except early flowers were eliminated from the inflorescence, these flowers produced more fruits and more seeds than the early flowers of control inflorescences, providing further support for the resource competition hypothesis and suggesting that the earliest fruits (even though they act as resource sinks) would perform even better in the absence of resource limitation.
The number of seeds per fruit and seed-to-ovule ratio did not always show a significant increase when only three early flowers were removed from the inflorescence. Even when two-thirds of the flowers on the inflorescence were removed and abundant pollen was supplied to the remaining flowers, only ~30% of ovules produced seed. This suggests that reallocation of resources within inflorescences of P. maritimum occurs largely during the fruit production stage. However, the number of seeds produced per fruit and the proportion of ovules producing seeds may be simultaneously subject to other restrictions that cause low seed production per fruit even when available resources increase.
A number of selective advantages of producing excess flowers, initiated fruits, and/or ovules have been postulated. First, the excess may allow for the quality of fruits and/or seeds to be improved by selective abortion (see above). Second, the excess may act as an insurance against variability in resource availability (including pollinators) and/or herbivore pressure. Third, flowers may achieve reproductive success largely through male function, despite being morphologically hermaphrodite. [For additional explanations see, for example, Stephenson, 1981
; Bawa and Webb, 1984
; Holtsford, 1985
; Nakamura, 1986
; Sutherland, 1987
; Lee, 1988
; Ehrlén, 1991, 1993
; Guitián, 1993, 1994.]
Burd (1994)
has suggested that, in species with "stochastic pollination," the optimum number of ovules per flower is frequently greater than the mean number of pollen tubes that can successfully develop. In P. maritimum, the unpredictability of pollen receipt (possibly related to the scarcity of floral visitors) may have contributed to the evolution of excess ovules and flowers.
An adequate understanding of among-flower or among-inflorescence patterns of variation in hermaphroditic plants requires consideration of male as well as female function, and investigation of whether the flowers of the species in question show any form of sexual specialization (see, e.g., Stephenson, 1981
; Bawa and Webb, 1984
; Brunet, 1996
; and references therein). In the case of Aquilegia caerulea, Brunet (1996) suggests that the low fruit set and low seed production of late-opening flowers is partly attributable to these flowers being both morphologically and functionally specialized for pollen production and dissemination. Specialization of late-opening and/or distal flowers for male function has also been reported in other species (e.g., Calochortus leichtlinii: Holtsford, 1985; Solanum carolinense: Solomon, 1988
; Myrosmodes cochleare: Berry and Calvo, 1991
; Solanum hirtum: Diggle, 1991, 1994
). In the present study, we did not take this possibility into account at the experimental design stage, so we are unable to rule out the possibility of specialization of late flowers for male function. However, the fact that potential fruit and seed production by late flowers were similar to that of earlier flowers argues against the possibility of sexual specialization. Nevertheless, we have occasionally found late flowers in which the style is absent or atrophied and that lack viable ovules, suggesting that there may be some degree of specialization for male function (M. Medrano, unpublished observations).
In conclusion, the low fruit and seed production by "late" flowers of inflorescences of P. maritimum under natural conditions cannot be attributed to insufficient pollen receipt or to architectural constraints. The principal proximate cause of the decreasing pattern of female reproductive success within inflorescences appears to be competition for resources among the flowers of each inflorescence, with "early" flowers sequestering more resources than "late" flowers.
FOOTNOTES
1 Manuscript received (??);revision accepted (??)
The authors thank María Cortizo and Manuel Martínez for their generous hospitality during the fieldwork, and the personnel of the Corrubedo Dunes Complex Natural Park for their assistance throughout the study; Pamela Diggle, Luis Navarro, José María Sánchez, Guy Norman, and two anonymous reviewers for comments that improved the manuscript; and Luis Navarro, Marisa Buide and Beatriz Pías for help at various stages in this study. The work was partially financed by the University of Santiago de Compostela, the Spanish Government's Dirección General de Enseñanza Superior (PB96-0856), and the Galician Government's Dirección General de Universidades (XUGA-10303B96). 
2 Author for correspondence (phone: 34-981-563100-14975; FAX: 34-981-594912; e-mail: bvmonime{at}usc.es
).
LITERATURE CITED
Bawa, K. S., and C. J. Webb. 1984 Flower, fruit and seed abortion in tropical forest trees: implications for the evolution of paternal and maternal reproductive patterns. American Journal of Botany 71: 736–751. [CrossRef][ISI]
Berry, P. E., and R. N. Calvo. 1991 Pollinator limitation and position dependent fruit set in the high Andean orchid Myrosmodes cochleare (Orchidaceae). Plant Systematics and Evolution 174: 93–101. [CrossRef][ISI]
Brunet, J. 1996 Male reproductive success and variation in fruit and seed set in Aquilegia caerulea (Ranunculaceae). Ecology 77: 2458–2471. [CrossRef][ISI]
———, and D. Charlesworth. 1995 Floral sex allocation in sequentially blooming plants. Evolution 49: 70–79. [CrossRef][ISI]
Burd, M. 1994 Bateman's principle and plant reproduction: the role of pollen limitation in fruit and seed set. Botanical Review 60: 83–139.
Carballeira, A., C. Devesa, R. Retuerto, E. Santillán, and F. Ucieda. 1983 Bioclimatología de Galicia. Fundación Pedro Barrié de la Maza, La Coruña, Spain.
Devlin, B. 1989 Components of seed and pollen yield of Lobelia cardinalis: variation and correlations. American Journal of Botany 76: 204–214. [CrossRef][ISI]
Diggle, P. K. 1991 Labile sex expression of andromonoecious Solanum hirtum: pattern of variation in floral structure. Canadian Journal of Botany 80: 967–973.
———. 1992 Development and the evolution of plant reproductive characters. In R. Wyatt [ed.], Ecology and evolution of plant reproduction: new approaches, 326–335. Chapman and Hall, New York, New York, USA.
———. 1994 The expression of andromonoecy in Solanum hirtum (Solanaceae): phenotypic plasticity and ontogenetic contingency. American Journal of Botany 81: 1354–1365. [CrossRef][ISI]
———. 1995 Architectural effects and the interpretation of patterns of fruit and seed development. Annual Review of Ecology and Systematics 26: 531–532. [CrossRef][ISI]
———. 1997 Ontogenetic contingency and floral morphology: the effects of architecture and resource limitation. International Journal of Plant Sciences 158 (supplement): S99-S107.
Ehrlén, J. 1991 Why do plants produce surplus flowers? A reserve-ovary model. American Naturalist 138: 918–933. [CrossRef][ISI]
———. 1992 Proximate limits to seed production in a herbaceous perennial legume, Lathyrus vernus. Ecology 73: 1820–1831.
———. 1993 Ultimate functions of non-fruiting flowers in Lathyrus vernus. Oikos 68: 45–52.
Emms, S. K. 1996 Temporal patterns of seed set and fitness returns on female allocation in Zigadenus paniculatus (Liliaceae), an andromonoecious lily. American Journal of Botany 83: 304–315. [CrossRef][ISI]
Goldingay, R. L., and R. J. Whelan. 1993 The influence of pollinators on fruit positioning in the Australian shrub Telopea speciosissima (Proteaceae). Oikos 68: 501–509. [CrossRef][ISI]
Guitián, J. 1993 Why Prunus mahaleb (Rosaceae) produces more flowers than fruits. American Journal of Botany 80: 1305–1309. [CrossRef][ISI]
———. 1994 Selective fruit abortion in Prunus mahaleb (Rosaceae). American Journal of Botany 81: 1555–1558. [CrossRef][ISI]
———, and L. Navarro. 1996 Allocation of reproductive resources within inflorescences of Petrocoptis grandiflora (Caryophyllaceae). Canadian Journal of Botany 74: 1482–1486.
Herrera, J. 1991 Allocation of reproductive resources within and among inflorescences of Lavandula stoechas (Lamiaceae). American Journal of Botany 78: 789–794. [CrossRef][ISI]
Holtsford, T. P. 1985 Nonfruiting hermaphroditic flowers of Calochortus leichtlinii (Liliaceae): potential reproductive functions. American Journal of Botany 72: 1687–1694. [CrossRef][ISI]
Karoly, K. 1992 Pollinator limitation in the facultatively autogamous annual, Lupinus nanus (Leguminosae). American Journal of Botany 79: 49–56. [CrossRef][ISI]
Lee, T. D. 1988 Patterns of fruit and seed production. In J. Lovett Doust and L. Lovett Doust [eds.], Plant reproductive ecology: patterns and strategies, 179–202. Oxford University Press, New York, New York, USA.
McKone, M. J. 1985 Reproductive biology of several bromegrasses (Bromus): breeding system, pattern of fruit maturation, and seed set. American Journal of Botany 72: 1334–1339. [CrossRef][ISI]
Medrano, M., P. Guitián, and J. Guitián. 1999 Breeding system and temporal variation in fecundity of Pancratium maritimum L. (Amaryllidaceae). Flora 194: 13–19. [ISI]
Méndez, M. 1997 Sources of variation in seed mass in Arum italicum. International Journal of Plant Sciences 158: 298–305.
Muñoz, A., and J. A. Devesa. 1987 Contribución al conocimiento de la biología floral del género Lavandula. II. Lavandula stoechas L. subsp. stoechas. Anales del Jardín Botánico de Madrid 44: 63–78.
Nakamura, R. R. 1986 Maternal investment and fruit abortion in Phaseolus vulgaris. American Journal of Botany 73: 1049–1057.
Navarro, L. 1996 Fruit-set and seed weight variation in Anthyllis vulneraria subsp. vulgaris (Fabaceae). Plant Systematics and Evolution 201: 139–148. [CrossRef][ISI]
Nilsson, L. A. 1983 Anthecology of Orchis mascula (Orchidaceae). Nordic Journal of Botany 3: 157–179.
Obeso, J. R. 1993 Seed mass variation in the perennial herb Asphodelus albus: sources of variation and position effect. Oecologia 93: 571–575. [CrossRef][ISI]
SAS. 1992 SAS user's guide: statistics. SAS Institute Inc., Cary, North Carolina, USA.
Schoen, D. J., and M. Dubuc. 1990 The evolution of inflorescence size and number: a gamete-packaging strategy in plants. American Naturalist 135: 841–857. [CrossRef][ISI]
Sokal, R. R., and F. J. Rohlf. 1995 Biometry, 3rd ed. W. H. Freeman, New York, New York, USA.
Solomon, B. P. 1988 Patterns of pre- and postfertilization resource allocation within an inflorescence: evidence for interovary competition. American Journal of Botany 75: 1074–1079. [CrossRef][ISI]
Stephenson, A. G. 1979 An evolutionary examination of the floral display of Catalpa speciosa (Bignoniaceae). Evolution 33: 1200–1209. [CrossRef][ISI]
———. 1980 Fruit set, herbivory, fruit reduction and fruiting strategy of Catalpa speciosa (Bignoniaceae). Ecology 61: 57–64. [CrossRef][ISI]
———. 1981 Flower and fruit abortion: proximate causes and ultimate functions. Annual Review of Ecology and Systematics 12: 253–279.
Sutherland, S. 1987 Why hermaphroditic plants produce many more flowers than fruits: experimental tests with Agave mckelveyana. Evolution 41: 750–759.
Tamas, I. A., D. H. Wallace, P. M. Ludford, and J. L. Ozbun. 1979 Effect of older fruits on abortion and abscisic acid concentration of younger fruits in Phaseolus vulgaris L. Plant Physiology 64: 620–622.
Thomson, J. D. 1985 Pollination and seed set in Diervilla lonicera (Caprifoliaceae): temporal patterns of flower and ovule development. American Journal of Botany 72: 737–740. [CrossRef][ISI]
———. 1989 Deployment of ovules and pollen among flowers within inflorescences. Evolutionary Trends in Plants 3: 65–68.
Udovic, D., and C. Aker. 1981 Fruit abortion and the regulation of fruit number in Yucca whipplei. Oecologia 49: 245–248.
Van Stevenick, R. F. M. 1957 Factors affecting the abscission of reproductive organs in yellow lupins (Lupinus luteus L.). I. The effect of different patterns of flower removal. Journal of Experimental Botany 8: 373–381.
Vaughton, G. 1993 Nonrandom patterns of fruit set in Banksia spinulosa (Proteaceae): interovary competition within and among inflorescences. International Journal of Plant Sciences 154: 306–313. [CrossRef]
———, and M. Ramsey. 1997 Seed mass variation in the shrub Banksia spinulosa (Proteaceae): resource constraints and pollen source effects. International Journal of Plant Sciences 158: 424–431. [CrossRef]
Wolfe, L. M. 1992 Why does the size of reproductive structures decline through time in Hydrophyllum appendiculatum (Hydrophyllaceae)?: developmental constraints vs. resource limitation. American Journal of Botany 79: 1286–1290. [CrossRef][ISI]
Wyatt, R. 1980 The reproductive biology of Asclepias tuberosa. I. Flower number, arrangement, and fruit set. New Phytologist 85: 119–131. [CrossRef][ISI]
———. 1982 Inflorescence architecture: how flower number, arrangement, and phenology affect pollination and fruit-set. American Journal of Botany 69: 585–594. [CrossRef][ISI]
Zar, J. D. 1996 Biostatistical analysis, 3rd ed. Prentice-Hall, Englewood Cliffs, New Jersey, USA.
Zimmerman, J. K., and T. M. Aide. 1989 Patterns of fruit production in a neotropical orchid: pollinator vs. resource limitation. American Journal of Botany 76: 67–73. [CrossRef][ISI]
This article has been cited by other articles:
|
|
 |
|
  M{a} L. Buide Disentangling the causes of intrainflorescence variation in floral traits and fecundity in the hermaphrodite Silene acutifolia Am. J. Botany, April 1, 2008; 95(4): 490 - 497. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. L. Marshall, J. Reynolds, N. J. Abrahamson, H. L. Simpson, M. G. Barnes, J. S. Medeiros, S. Walsh, D. M. Oliveras, and J. J. Avritt Do differences in plant and flower age change mating patterns and alter offspring fitness in Raphanus sativus (Brassicaceae)? Am. J. Botany, March 1, 2007; 94(3): 409 - 418. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. O. Brown and J. N. McNeil Fruit production in cranberry (Ericaceae: Vaccinium macrocarpon): a bet-hedging strategy to optimize reproductive effort Am. J. Botany, June 1, 2006; 93(6): 910 - 916. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. GUITIAN, M. MEDRANO, and J. E. OTI Variation in Floral Sex Allocation in Polygonatum odoratum (Liliaceae) Ann. Bot., September 1, 2004; 94(3): 433 - 440. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. L. BUIDE Intra-inflorescence Variation in Floral Traits and Reproductive Success of the Hermaphrodite Silene acutifolia Ann. Bot., September 1, 2004; 94(3): 441 - 448. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. P. Taylor and R. L. Obendorf Quantitative Assessment of Some Factors Limiting Seed Set in Buckwheat Crop Sci., November 1, 2001; 41(6): 1792 - 1799. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. J. Mazer and K. A. Dawson Size-dependent sex allocation within flowers of the annual herb Clarkia unguiculata (Onagraceae): ontogenetic and among-plant variation Am. J. Botany, May 1, 2001; 88(5): 819 - 831. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||
