FEC_580.fm Page 782 Monday, November 12, 2001 6:26 PM

Functional Ecology 2001 15, 782 – 790

Why do flowers of a hummingbird-pollinated mistletoe face down?

Blackwell Science Ltd

M. TADEY and M. A. AIZEN† Laboratorio Ecotono, CRUB, Universidad Nacional del Comahue, Unidad Postal Universidad, 8400 Bariloche, Río Negro, Argentina

Summary 1. Pendant flowers are common among hummingbird-pollinated plants. A downward orientation of the flower or inflorescence could represent an adaptation to avoid either flower flooding or direct pollen losses from anthers or stigmas under rainy conditions. 2. We studied the adaptive significance of this trait experimentally in Tristerix corymbosus Kuijt (Loranthaceae), a mistletoe native to the temperate forests of southern South America. We applied three treatments: (i) natural pendant inflorescences; (ii) inflorescences tethered to face up; and (iii) inflorescences tethered to face down (as a control for tethering). We also considered natural exposure to rain as a second factor. 3. The treatments did not differ significantly in either nectar volume or concentration. Flowers exposed to rain for most of their lives contained more diluted nectar than those that remained dry, but this result did not depend on either inflorescence or flower orientation. 4. We found significantly fewer pollen tubes in styles of flowers from inflorescences tethered to face up than in flowers receiving the other two treatments, but this could not be attributed to a direct effect of rain exposure. Inflorescence orientation did not affect either the number of pollen grains left in anthers or seed set. No strong evidence was found for differential visitation by hummingbirds in relation to a flower’s angle. 5. The results of this work support neither the flower-flooding nor the pollen-protection hypothesis. However, a flower’s orientation may affect the extent of within-flower selfpollination or the efficiency of pollen transfer from a hummingbird’s bill onto a flower’s stigma. Key-words: Female and male reproductive success, flower orientation, hummingbird pollination, pollen tubes, rain Functional Ecology (2001) 15, 782 – 790

Introduction Flowering plants that depend on animal agents of pollination may ensure successful reproduction through traits that increase attraction and effective visitation by pollinators (Bertin 1982; Howe & Lynn 1988; Vaknin, Tov & Eisikowitch 1996; Waser 1983). These traits can affect plant fitness through pollen receipt and seed set (female function), and pollen donation and seed siring (male function). Several studies have revealed that slight variations in size, shape or colour can affect visitation to individual flowers, thus affecting female as well as male functions (Campbell 1989; Dafni & Kevan 1997; Meléndez-Ackerman, Campbell & Waser 1997; Murcia 1990; Podolsky 1992; Podolsky 1993; Temeles 1996). In addition to these traits, flower angle in relation to the horizontal may also affect pollination and plant © 2001 British Ecological Society

†Author to whom correspondence should be addressed. E-mail: [email protected]

reproductive success. For example, species with hanging flowers and tubular corollas presumably avoid flower flooding during rain, exposing the external side of petals instead of the inner part of the corolla. Consequently they avoid nectar dilution which would affect visitation by discriminating pollinators (Sprengel 1793 translated in Lloyd & Barrett 1996). Another possible advantage of down-facing flowers could be to protect pollen from being washed from anthers or stigmas: petals act as ‘umbrellas’, enhancing both female and male fitness (Broyles & Wyatt 1990; Campbell 1989; Devlin, Clegg & Ellstrand 1992; Dudash 1991; Galen 1992; Schoen & Stewart 1986). Although facing down is a widespread flower trait among angiosperms, it may represent an important adaptation to vertebrate pollinators that remain active during rainy or even snowy weather (M.A.A., unpublished results). Pendant flowers or inflorescences appear to occur frequently among hummingbird-pollinated plants. This trait, plus red coloration, tubular shape, 782

FEC_580.fm Page 783 Monday, November 12, 2001 6:26 PM

783 Adaptive significance of flower orientation

absence of odour, and diluted nectar rich in sucrose are commonly associated in the so-called ‘hummingbird pollination syndrome’ (Grant & Grant 1968; van der Pijl 1961; Proctor, Yeo & Lack 1996). Despite its potential importance for plant fitness, however, flower orientation has received much less attention than other morphological traits of hummingbirdpollinated plants, such as flower colour, size or shape (Campbell 1989; Campbell, Waser & MeléndezAckerman 1997; Feinsinger & Busby 1987; Fenster 1991; Hurlbert et al. 1996; Meléndez-Ackerman 1997; Murcia & Feinsinger 1996; Podolsky 1992; Podolsky 1993; Straw 1956; Stanton, Snow & Handel 1986; Temeles 1996). Here we assess the adaptive significance of a downfacing orientation by altering natural flower angle in the hanging inflorescences of the hemiparasite Tristerix corymbosus Kuijt (Loranthaceae), a hummingbirdpollinated plant native to the southern Andes. In particular, we tested the flower-flooding and pollenprotection hypotheses detailed above, by examining simultaneously how flower angle and exposure to rain affect nectar quality and quantity, pollinator visitation, pollen in anthers, pollination levels, and seed set.

Materials and methods     Tristerix corymbosus, the ‘quintral’, occurs along the western rim of southern South America in Chile and Argentina, from 32° S to 42° S. It inhabits regions with >4000 mm of annual rainfall to the south in the

Valdivian rainforest, to <500 mm to the north in the Mediterranean region of central Chile (Kuijt 1988). Throughout its geographical range, it parasitizes more than 20 host species. Tristerix corymbosus has a very extended flowering phenology. In our study area, flowering lasts nearly 9 months from the beginning of March to the end of November (Ruffini 1992; M.A.A., unpublished results). Well established plants can produce several hundred flower buds. Flowers are arranged in short-peduncled racemes or corymbs, with four to 14 flowers. Inflorescences are usually pendant, or more rarely erect with flowers spreading in different angles. Flowers are red, symmetrical, tubular, and 4– 5 cm long × 0·3 cm wide. The four petals, not truly fused, reflex during anthesis exposing four yellow filaments and a capitate style which protrudes about 0·5 cm beyond the anthers (Fig. 1a). The filaments and style turn red when a flower senesces. Despite self-compatibility, pollen transfer and seed set in T. corymbosus are mostly pollinator-mediated (M.A.A., unpublished results). In our study area, the exclusive pollinator of this plant species throughout most of its flowering phenology (from March to October) is Sephanoides sephaniodes, a trochilid hummingbird native to the temperate forests of southern South America (Ruffini 1992; Smith-Ramírez 1993). Pollen of T. corymbosus can frequently be found in both the dorsal and ventral surface of the hummingbird’s bill and head (Ruffini 1992). Successfully pollinated flowers mature into green viscous berries containing one naked seed (Amico & Aizen 2000). The study site was in the Parque Municipal LlaoLlao, about 30 km west of San Carlos de Bariloche, Argentina (41°8′ S, 71°19′ W). The annual precipitation is approximately 1800 mm and the mean temperature varies from 2 °C in winter to 13·6 °C in summer. Most precipitation (88%) falls between March and November – the flowering period of T. corymbosus – and snowstorms are common during winter (Barros et al. 1983). The study population of T. corymbosus was located in a large gap (≈100 × 50 m) near the path to Villa Tacul, amidst a mixed, old-growth forest dominated by Nothofagus dombeyi and the conifer Austrocedrus chilensis. The population consisted of about 20 large flowering individuals (60–180 inflorescences per plant) parasitizing branches of the shrubs Aristotelia maqui and Maytenus boaria, the two most common hosts in the study area.

 

Fig. 1. Diagrams showing (a) the developmental stages of a flower of Tristerix © 2001 British corymbosus (left to right: flower bud; 1-day-old flower; mature flower close to Ecological Society, senescence); and (b) the three inflorescence treatments. In (b) a solid line indicates a flower’s major axis;,α+ and α– examples of flowers showing positive and negative angles Functional Ecology with respect 15 , 782 – 790 to the horizontal (dotted line), respectively.

In April 1998, we randomly selected 10 individuals of T. corymbosus and applied one of the following treatments to individual inflorescences: (i) natural (control) pendant inflorescences (CO); (ii) pendant inflorescences tethered to face up (FU); and (iii) pendant inflorescences tethered to face down, as a control for tethering (FD). Pieces of thin copper wire tied around an inflorescence’s peduncle and the adjacent branch

FEC_580.fm Page 784 Monday, November 12, 2001 6:26 PM

784 M. Tadey & M. A. Aizen

© 2001 British Ecological Society, Functional Ecology, 15, 782 – 790

were used to tether inflorescences in the desired position (Fig. 1b). Each of the three treatments was applied to five randomly chosen inflorescences per plant (five inflorescences × three treatments = 15 treated inflorescences per plant). In each inflorescence selected, we marked five flower buds using numbered jewellery tags tied around the pedicel. Tags were hidden behind an inflorescence’s subtending leaves. The status of each tagged bud was checked daily or every other day, from April to the end of September when all marked flowers had senesced. On each sampling day tagged buds were classified as: bud; open flower; senescent flower; swollen ovary; developing fruit. At the end of the flowering season we continued sampling once every 2 weeks until late December when all marked fruits had matured. The fate of 750 flowers was followed. After anthesis, the angle of the major axis of each marked flower relative to the horizontal was recorded using a clinometer (AcuAngle, A-300, accuracy = 0·1°; Fig. 1b). We avoided any shaking of flowers that might cause artificial pollen removal or deposition. Angles ranged between –90° and +90°, with negative angles representing down-facing flowers and positive angles representing up-facing flowers. A horizontal flower had an angle of 0°. We estimated flower longevity as the number of days between the date a flower was first observed open and the date it was first recorded as senescent. To estimate pollination success, we collected styles of marked flowers 1–2 weeks after flower senescence. Styles were placed in individual microcentrifuge tubes and fixed in formalin : acetic acid : ethyl alcohol 5 : 5 : 90. Observations showed that style removal at this time did not affect future fruit development (M.A.A., unpublished results). In the laboratory, styles were cleared overnight in 10 N NaOH and stained with 0·1% aniline blue in 0·1  K3PO4 (Martin 1959), squashed, and examined with an epifluorescence microscope. We counted pollen tubes just below the stigma to estimate the number of germinated pollen grains. To determine whether hummingbirds prefer to visit a particular inflorescence type, we observed hummingbird visits to focal inflorescences for 2 h per sampling date. As we could not observe all the plants at once, we observed groups of two or three different plants for 30 min at a time. Given that we could record only a few visits to experimental inflorescences, we supplemented information on hummingbirds’ preferences by measuring the angles of visited flowers in non-experimental inflorescences over the whole sampling period. Each time we measured the angle of a visited flower, we also measured the angle of a randomly chosen flower in the nearest inflorescence within the same plant (cf. Stanton & Galen 1989). Because we could not measure nectar non-destructively, we collected non-tagged, fresh flowers (2–4 days old) from experimental inflorescences (n = 397 flowers). The flower angle was measured before collection. Nectar standing crop was extracted with 5 µl capillary tubes

and its concentration measured with a hand refractometer (Reichter, model 10431, Leica Inc., Buffalo, NY, USA). Sugar concentration in sucrose equivalents (100 × mg solute/ mg solution) was converted to mg solute µl–1 based on tabulated values in Kearns & Inouye (1993, p. 172) and combined with sample volumes to estimate nectar sugar content in mg. We also collected the anthers from each of these flowers before nectar measurement, and kept them in separated 0·5 mm microcentrifuge tubes containing 70% ethyl alcohol. The volume of each tube was raised to 0·5 ml by adding drops of detergent solution. After vortexing for 60 s, the number of pollen grains remaining per anther was estimated from two aliquots using a haemocytometer (Aizen & Raffaele 1996; Aizen & Raffaele 1998). To test whether inflorescence treatment had an effect on nectar production, we applied the same treatments as above (CO, FU, FD) to inflorescences of five plants that were not used for the other observations. Each treatment was applied to three inflorescences per plant (three inflorescences × three treatments = nine treated inflorescences per plant). For a given sampling date we collected one fresh, opened flower from each experimental inflorescence to measure the standing volume of nectar. We then marked another flower with a jewellery tag and bagged the focal inflorescences with nylon netting. Tagged flowers were collected 24 h later, and the volume and concentration of secreted nectar was measured as above. For each inflorescence we estimated nectar production per flower during a 24 h period by subtracting the standing crop of the flower collected just before bagging from the volume measured in the tagged flower. This procedure was repeated for each focal inflorescence at least twice during the flowering period. We classified each date of the flowering period sampled (April–September 1998) as ‘rainy’ or ‘dry’ based on the occurrence of rain or snow. To document precipitation, we used five glasses, 6 cm in diameter, exposed to the open sky as water collectors. The glasses were placed on the ground within the sampled gap at the beginning of the sampling period and checked on every sampling date. We supplemented information collected in the field with precipitation records from the Instituto Nacional de Technología Agropecuaria (INTA) meteorological station, situated 30 km east of the study site.

  To test whether inflorescence tethering effectively modified flower angles, we used a mixed-model  that included inflorescence treatment as the fixed factor and plant as the random factor. The five angle measurements for each inflorescence were averaged before analysis. We used the same  model to test for the effect of treatment on 24 h nectar production (volume and concentration) averaged over each experimental inflorescence. In this and the  model

FEC_580.fm Page 785 Monday, November 12, 2001 6:26 PM

785 Adaptive significance of flower orientation

factor interaction (treatment × rain × plant) from the final model because this term was never significant (P > 0·5 in all cases). Because our design was not completely balanced, we based significance tests on type III sums of squares (SAS 1988). Approximated error terms for each factor included in the model were estimated using the RANDOM statement in   (SAS 1988). We report visit frequencies, but did not analyse these statistically because we recorded only 15 visits to focal inflorescences. On the other hand, we recorded many visits to flowers in non-experimental inflorescences because of their comparatively greater abundance. Comparisons of angles between visited and paired randomly chosen flowers were analysed using a nonparametric Wilcoxon test for paired observations because the data were not normally distributed.

described below, the sum of squares associated with inflorescence treatment was decomposed in two a priori, orthogonal contrasts (Sokal & Rohlf 1981). These contrasts tested for the independent effects of inflorescence orientation (CO + FD vs FU), and inflorescence tethering (CO vs FD). We used a three-factor  model to analyse flower longevity; nectar standing crop (volume, concentration and total sugar content); number of pollen grains remaining per anther; number of pollen tubes per style; and fruit set (number of mature fruits/ number of flowers). This model assessed the effects of inflorescence treatment (fixed), plant (random), and exposure to rain (fixed). We averaged values for ‘wet’ and ‘dry’ flowers for each experimental inflorescence, and considered these means as individual observations (Mead 1988). For the analysis of flower longevity, pollen tubes and fruit set, a flower was classified as wet if it was exposed to precipitation for ≥50% of the days it stayed open, otherwise the flower was considered dry. For nectar variables and pollen remaining per anther, we considered a flower as wet if it was exposed to precipitation during the same day or the day before collection (and dry otherwise) because we did not know the exact date of anthesis of these flowers. These two classifications were comparable, because flowers collected fresh for nectar analysis were less than 2–4 days old. In addition, standing nectar might be influenced more strongly by precipitation occurring on the day of measurement or the day before. We tested for an effect of inflorescence orientation on nectar, pollination levels (estimated as number of pollen tubes), pollen removal and seed output through the treatment factor, particularly through the CO + FD vs FU contrast. In addition, we sought evidence for the flowerflooding and pollen-protection hypotheses by looking at the significance of the treatment × rain interaction, as the adaptive consequence of flower orientation should be more apparent for flowers exposed to rain. To increase statistical power, we dropped the three–

Results Inflorescence treatment strongly and significantly affected the orientation of individual flowers (CO + FD vs FU contrast, Table 1). A large proportion (54·8%) of flowers within inflorescences forced to face up exhibited positive angles, whereas in control and tethereddown inflorescences only 9·5 and 1·3% of all flowers, respectively, exhibited angles >0° (Fig. 2). Inflorescence tethering per se did not significantly affect flower orientation (CO vs FD contrast, Table 1). Because flower angles varied significantly among plants, the extent to which inflorescence treatment modified natural flower angles was also plant-dependent (treatment–plant interaction, Table 1). Inflorescence treatment did not affect flower longevity. In contrast, exposure to rain – significantly reduced flower life span (X ± 1 SE = 7·5 ± 0·2 vs 6·6 ± 0·2 days; Table 1). We recorded 15 hummingbird visits to experimental inflorescences. We did not conduct a formal statistical analysis, but there was no evidence of differential visitation in relation to inflorescence position: five visits were to control inflorescences, four visits were to

Table 1. Results of mixed-model s testing the effects of inflorescence treatment (fixed) and plant (random) on individual flower angle, and of these two factors plus exposure to rain (fixed) on flower longevity Flower angle (degrees)

Treatment CO + FD vs FU CO vs FD Rain Plant Treatment × rain Treatment × plant Rain × plant

© 2001 British Ecological Society, Functional Ecology, 15, 782 – 790

Flower longevity (days)

df†

F

df†

F

2,18 1,18 1,18 – 9,120 – 18,120 –

67·95*** 134·28*** 1·42 – 15·18*** – 3·41*** –

2,19 1,19 1,19 1,9·4 9,12·8 2,237 18,237 9,237

1·20 2·06 0·44 15·08** 0·60 0·02 0·55 0·62

Treatment effects were decomposed in two orthogonal contrasts: (1) control (CO) and inflorescences tethered to face down (FD) vs inflorescences tethered to face up (FU), and (2) CO vs FD. F-tests considered type III sums of squares. Approximated denominator sums of squares and degrees of freedom were estimated using the RANDOM statement in   (SAS 1988). †Numerator degrees of freedom, denominator degrees of freedom. **P < 0·01, ***P < 0·001.

FEC_580.fm Page 786 Monday, November 12, 2001 6:26 PM

786 M. Tadey & M. A. Aizen

Table 2. Results of mixed-model s testing the effects of inflorescence treatment (fixed), exposure to rain (fixed) and plant (random) on nectar standing crop (volume, sugar concentration and total sugar content)

Treatment CO + FD vs FU CO vs FD Rain Plant Treatment × rain Treatment × plant Rain × plant

Nectar volume (µl)

Sugar concentration (%)

Nectar sugar (mg)

df†

F

df†

F

df†

F

2,26·9 1,26·9 1,26·9 1,14·5 9,12·9 2,143 18,143 9,143

0·83 0·05 1·46 7·27* 2·25(*) 3·17* 1·77* 1·36

2,37·2 1,37·2 1,37·2 1,11·6 9,8·4 2,143 18,143 9,143

0·47 0·98 0·06 19·53*** 1·02 0·93 0·88 2·74**

2,28·9 1,28·9 1,28·9 1,15·7 9,10·8 2,143 18,143 9,143

0·76 0·24 0·27 23·97*** 3·10* 2·92(*) 1·47 1·13

Treatment effects were decomposed in two orthogonal contrasts: (1) control (CO) and inflorescences tethered to face down (FD) vs inflorescences tethered to face up (FU), and (2) CO vs FD. F-values considered type III sums of squares. Approximated denominator sums of squares and degrees of freedom were estimated using the RANDOM statement in   (SAS 1988). †Numerator degrees of freedom, denominator degrees of freedom. (*)0·05 < P < 0·10, *P < 0·05, **P < 0·01, ***P < 0·001.

Fig. 2. Frequency distributions of individual flower angles relative to the horizontal for control inflorescences (CO, dashed curve); tethered-down inflorescences (FD, dotted curve); and tethered-up inflorescences (FU, solid curve). For statistical details, see Table 1.

© 2001 British Ecological Society, Functional Ecology, 15, 782 – 790

inflorescences tethered upwards, and six to inflorescences tethered downwards. Comparisons of flower angles between visited and paired, randomly chosen flowers did not support the hypothesis of differential visitation to more negatively oriented flowers. Indeed, we found a trend in the opposite direction, as the mean angle (±1 SE) of visited flowers was –52° ± 3° whereas that of randomly chosen flowers was –56° ± 3° (T = 2469, Z = 2·15, n = 114, P = 0·03). Inflorescence treatment did not significantly affect volume, concentration or total sugar content of the nectar standing crop (Table 2, Fig. 3). Similarly, we found no significant association across inflorescence treatments and plants between individual flower angle and nectar standing volume (r = 0·003, n = 373, P = 0·95); sugar concentration (r = – 0·051, n = 372, P = 0·32); or total sugar content (r = – 0·019, n = 373, P = 0·71). Flowers exposed to rain contained smaller volumes of more diluted nectar than flowers not exposed to rain. Thus the standing nectar of wet flowers contained significantly less sugar than that of dry flowers (Table 2, Fig. 3). Differences in volume and

Fig. 3. Mean (± 1 SE) standing crop of nectar (volume, sugar concentration and total sugar content) in relation to inflorescence treatment and rain exposure. For statistical details, see Table 2.

sugar content of the standing nectar between wet and dry flowers tended to be larger for tethered-up inflorescences than for either control or tethered-down inflorescences (treatment–rain interaction, Table 2, Fig. 3).

FEC_580.fm Page 787 Monday, November 12, 2001 6:26 PM

787 Adaptive significance of flower orientation

Table 3. Results of mixed-model s testing the effects of inflorescence treatment (fixed), exposure to rain (fixed) and plant (random) on number of pollen grains remaining per anther, number of pollen tubes per style and fruit set No. grains /anther

Treatment CO + FD vs FU CO vs FD Rain Plant Treatment × rain Treatment × plant Rain × plant

No. tubes /style

No. fruits/no. flowers

df†

F

df†

F

df†

F

2,42·9 1,42·9 1,42·9 1,23 9,14 2,36 18,136 9,136

1·83 0·11 2·79 1·76 2·80* 0·97 0·67 0·54

2,18·4 1,18·4 1,18·4 1,9·2 9,7·2 2,238 18,238 9,238

3·81* 6·21* 1·57 35·06*** 6·56** 0·77 1·09 1·20

2,16·2 1,16·2 1,16·2 1,8·1 8,9·1 2,218 16,218 8,218

0·47 0·53 0·43 1·20 0·17 0·67 1·47 1·36

The treatment factor was decomposed in two orthogonal contrasts: (1) control (CO) and inflorescences tethered to face down (FD) vs inflorescences tethered to face up (FU), and (2) CO vs FD. F-tests considered Type III sums of squares. Approximated denominator sums of squares and degrees of freedom were estimated using the RANDOM statement in   (SAS 1988). †Numerator degrees of freedom, denominator degrees of freedom. *P < 0·05, **P < 0·01, ***P < 0·001.

Neither inflorescence treatment nor exposure to rain affected the number of pollen grains remaining in a flower’s anthers. In contrast, both inflorescence treatment and exposure to rain significantly influenced pollen receipt (Table 3, Fig. 4). Inflorescence orientation significantly affected the mean number of pollen tubes counted in a flower’s style (CO + FD vs FU contrast, Table 3). Flowers in tethered-up inflorescences had 25 and 15% fewer pollen tubes compared to control and tethered-down inflorescences, respectively (Fig. 4). Similarly, pollen receipt varied negatively with individual flower angle (r = – 0·113, n = 659, P < 0·005). Tethering per se did not affect pollen receipt (CO vs FD contrast, Table 3). More important than inflorescence orientation was the effect of rain exposure on a flower’s pollination. Flowers exposed to rain during more than half their life span had 40% fewer pollen tubes in their styles than less-exposed flowers (Table 3, Fig. 4). However, the effects of inflorescence treatment or rain exposure on pollination levels were not large enough to influence final female reproductive success, as neither factor significantly affected fruit set (Table 3, Fig. 4).

Discussion

Fig. 4. Mean (± 1 SE) number of pollen grains remaining per anther, number of pollen tubes per style, and fruit set (number flowers/number fruits) in relation to inflorescence treatment and rain exposure. For statistical details, see Table 3.

© 2001 British Ecological Society, Functional Ecology, 15, 782 – 790

Inflorescence treatment did not significantly affect either the volume or concentration of the nectar produced after 24 h (F2,8 = 0·33, P = 0·72 for nectar volume; F2,8 = 0·42, P = 0·54 for sugar concentration).

Neither overall inflorescence orientation nor individual flower angles affected properties of the standing nectar in Tristerix corymbosus. On the other hand, nectar found in flowers on rainy days was, on average, more diluted than on non-rainy days. Lower nectar concentration did not occur because of rain pouring into either down-facing or up-facing flowers. Contrary to expectations, flowers open during rain had significantly smaller nectar volumes than flowers that were not exposed to rain during or just before nectar sampling. In particular, the smallest nectar volumes were sampled during or immediately after rain in flowers that were in tethered-up inflorescences. In many plant species, nectar secretion is an active metabolic process influenced by temperature (Aizen & Basilio 1998;

FEC_580.fm Page 788 Monday, November 12, 2001 6:26 PM

788 M. Tadey & M. A. Aizen

© 2001 British Ecological Society, Functional Ecology, 15, 782 – 790

Pleasants & Chaplin 1983; Rathcke 1992). The differential decrease in the volume of standing nectar observed in flowers exposed to rain may have resulted from a slower metabolism, perhaps in association with cooler daytime temperatures that characterize cloudy days. In addition, flowers in tethered-up inflorescences spread in a wider range of angles than in either control or down-facing inflorescences (Fig. 2). This greater spread, combined with an up-facing orientation, may relate to less flower insulation and thus to more exposure of flowers to temperature fluctuations. This may also explain why we sampled the highest nectar volumes in flowers that were open during mostly dry, sunny weather in tethered-up inflorescences (Fig. 3). Taken together, these results refute the flowerflooding hypothesis. This conclusion was confirmed by a straightforward experiment. Tristerix corymbosus flowers are not flooded even when submerged under running tap water (M.T. and C. Brewer, unpublished results). The narrow, tubular shape, combined with the hydrophobic and hydrophilic properties of the inner and outer petals’ surfaces, may be associated with this strong water repellence (C. Brewer, personal communication). The importance of surface properties for water retention has been studied in leaves (Brewer & Smith 1995; Brewer & Smith 1997), but we do not know of comparable studies conducted with flowers. Hummingbirds are highly discriminating pollinators that can distinguish between plants, between inflorescences within plants and even between flowers within inflorescences, based on nectar content (Campbell et al. 1997; Meléndez-Ackerman 1997; Sutherland & Gass 1995). As nectar quality was not affected by either inflorescence orientation or individual flower angle in our study, it is not surprising that we found no evidence of differential visitation of down-facing inflorescences or individual flowers. Indeed, we recorded a slight tendency for hummingbirds to visit less pendant flowers than those we chose at random. This might relate to a stronger visual attraction and greater accessibility of less pendant flowers, which usually occur in the outer part of T. corymbosus inflorescences. Even though hummingbirds did not discriminate against up-facing inflorescences, we recorded fewer pollen tubes in flowers collected from those inflorescences than from flowers in pendant or tethered-down inflorescences. As T. corymbosus plants present several open flowers at the same time, and stigma receptivity overlaps anther dehiscence within flowers (M.A.A., personal observation), many of the grains deposited on the stigmas could come from the same plant or even the same flower. For instance, Bertin (1982) found that around 50% of the pollen grains deposited on the stigma of hummingbird-pollinated Campsis radicans came from the same plant due to a combination of geitonogamous (between flowers of the same plant) and autogamous (within-flower) pollination (see also de Jong et al. 1992). A bagging experiment conducted in T. corymbosus showed that about 20 – 30% of the

pollen tubes growing in the style came from pollen of the same flower (M.A.A., unpublished results). Chances for autogamous pollination may vary with flower angle, and may explain why we found less pollination in upfacing inflorescences. In erect flowers of T. corymbosus, reduced chances for self-pollen deposition may be related to the fact that, unlike pendant flowers, the style protrudes upwards beyond the anthers. Hence the pollen that falls down due to gravity cannot reach the stigma (Motten & Antonovics 1992). An alternative but non-exclusive explanation for reduced pollination of up-facing inflorescences involves the way hummingbirds visit flowers (Campbell, Waser & Price 1994; Hurlbert et al. 1996; Murcia & Feinsinger 1996). When a hummingbird visits a pendant flower, it may make a more consistent contact between its head and either anthers or style, due to its vertical movement when hovering while feeding on nectar. We observed that hummingbirds approached pendant flowers of T. corymbosus from below and contacted sexual parts frequently while pushing through a flower’s entrance in search of nectar. On the few occasions we could observe hummingbirds visiting erect flowers, they approached flowers from above or the side, sometimes missing or reducing contact with either anthers or stigma. In an experimental study of Impatiens capensis flowers, Hurlbert et al. (1996) found that floral mobility, a trait often associated with facing down, increased pollen deposition on a hummingbird’s bill and crown by increasing handling times. Although we did not measure the cost of flower manipulation imposed on the hummingbird S. sephaniodes, a down-facing orientation of T. corymbosus flowers might increase handling times, resulting in greater stigmatic pollen deposition. In addition to an effect of inflorescence orientation, we found that rain strongly affected pollination. Flowers open during rainy weather for most of their life had fewer pollen tubes growing in the style than flowers that were open during mostly dry conditions. The reduced longevity or lower nectar volumes and concentrations found in T. corymbosus flowers exposed to rain might have diminished the number of visits that a flower received over its lifespan. Rain might also enhance pollen detachment from the stigma, or from a hummingbird’s bill, or may simply make pollen deposition more difficult (Bertin & Sholes 1993; Corbet 1990). In any event, flower orientation does not modify the reduction in pollen loads by rainfall, as this factor affected flowers similarly in the three types of inflorescences. Although flower angle affected pollen deposition, perhaps by changing the efficiency of pollen transfer, it did not affect the number of pollen grains remaining in flowers of T. corymbosus. A caveat should be introduced here. Given that flower age strongly determines the amount of pollen remaining in the anthers (Ashman & Schoen 1994; Galen 1992; Murcia 1990), the capacity to detect a significant effect of inflorescence

FEC_580.fm Page 789 Monday, November 12, 2001 6:26 PM

789 Adaptive significance of flower orientation

treatment on this variable might be impaired by not controlling for flower age either experimentally or statistically. Although we did not precisely control the age of the flowers collected for counting pollen (2–4 days old), Fig. 4 reveals no trend for up-facing inflorescences to lose more pollen than the other two types of control inflorescences. We did not find any trend for flowers exposed to rain to retain fewer pollen grains in the anthers than flowers that were open during mostly dry conditions. Independent of rain, flower orientation could affect male function if this trait influences the positioning of pollen removed by a hummingbird and thus its ability to ‘father’ seeds on the same or other plants (cf. Murcia & Feinsinger 1996; Waser 1983). This possibility remains to be tested. Taking these results together, we found no support for the hypothesis that a down-facing flower protects sexual parts better against direct pollen losses from the anthers or stigma under rainy conditions. The pendant flower of T. corymbosus is not an efficient umbrella. We also found no convincing evidence that inflorescence orientation affects fruit production. The fact that more pollination in pendant inflorescences did not translate into greater fruit set suggests that this effect was not strong enough to limit pollination. In a parallel study we found that natural pollination levels would have to be reduced by >50% to affect fruit set (M.A.A., unpublished results). However, we cannot rule out the possibility that changes in pollen loads of 10 – 30%, such as we found, might affect seed quality rather than seed quantity (Lee 1984; Lee 1988; Marshall & Folsom, 1991; Mulcahy 1979), although these effects have been rarely recorded in nature. Even though many hummingbird-pollinated species produce pendant flowers and are very abundant in wet and cloudy habitats throughout the Neotropics (Bawa 1990; Sazima, Buzato & Sazima 1996; Stiles 1981), we cannot conclude from our experimental study that facing down represents an adaptation to vertebrate pollinators, such as hummingbirds, that remain active under rainy conditions. More experimental and comparative research in a variety of systems is needed to determine whether this widespread floral trait plays any role in protecting flowers from rain.

Acknowledgements

© 2001 British Ecological Society, Functional Ecology, 15, 782 – 790

We are particularly grateful to Lawrence D. Harder for his careful editing and useful suggestions, and Carole Brewer and Thomas Thompson for fruitful discussions and for helping us to set the experiment. We also thank Diego P. Vázquez, Javier Puntieri, Nickolas M. Waser, and an anonymous reviewer for their valuable comments on an earlier version of this manuscript. This research was supported by the National Geographic Society (Grant no. 6192-98) and by CONICET, the National Research Council of Argentina (PEI no. 0018/97).

References Aizen, M.A. & Basilio, A. (1998) Sex differential nectar secretion in protandrous Alstroemeria aurea (Alstroemeriaceae): is production altered by pollen removal and receipt? American Journal of Botany 85, 245–252. Aizen, M.A. & Raffaele, E. (1996) Nectar production and pollination in Alstroemeria aurea: responses to levels and patterns of flowering shoot defoliation. Oikos 76, 312–322. Aizen, M.A. & Raffaele, E. (1998) Flowering-shoot defoliation affects pollen grain size and postpollination pollen performance in Alstroemeria aurea. Ecology 79, 2133–2142. Amico, G. & Aizen, M.A. (2000) Mistletoe seed dispersal by a marsupial. Nature 408, 929–930. Ashman, T.L. & Schoen, D.J. (1994) How long should flowers live? Nature 371, 788 –790. Barros, V., Cordon, V., Moyano, C., Mendez, R., Forquera, J. & Pizzio, O. (1983) Cartas de Precipitación de la Zona Oeste de las Provincias de Río Negro y Neuquén. Facultad de Ciencias Agrarias, Universidad Nacional del Comahue, Cinco Saltos, Argentina. Bawa, K.S. (1990) Plant–pollinator interactions in tropical forests. Annual Review of Ecology and Systematics 21, 399–422. Bertin, R.I. (1982) Floral biology, hummingbird pollination and fruit production of trumpet creeper (Campsis radicans, Bignoniaceae). American Journal of Botany 69, 122–134. Bertin, R.I. & Sholes, O.D.V. (1993) Weather, pollination and the phenology of Geranium maculatum. American Midland Naturalist 129, 52 – 66. Brewer, C.A. & Smith, W.K. (1995) Leaf surface wetness and gas exchange in the pond lily Nuphar polysepalum (Nymphaeaceae). American Journal of Botany 82, 1271–1277. Brewer, C.A. & Smith, W.K. (1997) Patterns of leaf surface wetness for montane and subalpine plants. Plant, Cell and Environment 20, 1–11. Broyles, S. & Wyatt, R. (1990) Paternity analysis in a natural population of Asclepias exaltata: multiple paternity, functional gender, and the ‘pollen donation hypothesis’. Evolution 44, 1454 –1468. Campbell, D.R. (1989) Measurement of selection in a hermaphroditic plant: variation in male and female reproductive success. Evolution 43, 318–334. Campbell, D.R., Waser, N.M. & Price, M.V. (1994) Indirect selection of stigma position in Ipomopsis aggregata via a genetically correlated trait. Evolution 48, 55–68. Campbell, D.R., Waser, N.M. & Meléndez-Ackerman, E.J. (1997) Analyzing pollinator-mediated selection in a plant hybrid zone: hummingbird visitation patterns on three spatial scales. American Naturalist 149, 295–315. Corbet, S.A. (1990) Pollination and the weather. Israel Journal of Botany 39, 13 – 30. Dafni, A. & Kevan, P.G. (1997) Flower size and shape: implications in pollination. Israel Journal of Plant Sciences 45, 201– 211. Devlin, B., Clegg, J. & Ellstrand, N.C. (1992) The effect of flower production on male reproductive success in wild radish populations. Evolution 46, 1030–1042. Dudash, M.R. (1991) Plant size effects on female and male function in hermaphroditic Sabatia angularis (Gentianaceae). Ecology 72, 1004 –1012. Feinsinger, P. & Busby, W. (1987) Pollen carryover: experimental comparisons between morphs of Palicourea lasiorrachis (Rubiaceae), a distylous, bird-pollinated, tropical treelet. Oecologia 73, 231–235. Fenster, C.B. (1991) Selection on floral morphology by hummingbirds. Biotropica 23, 98–101. Galen, C. (1992) Pollen dispersal dynamics in an alpine wild flower, Polemonium viscosum. Evolution 46, 1043–1041. Grant, K.A. & Grant, V. (1968) Hummingbirds and their Flowers. Columbia University Press, New York.

FEC_580.fm Page 790 Monday, November 12, 2001 6:26 PM

790 M. Tadey & M. A. Aizen

© 2001 British Ecological Society, Functional Ecology, 15, 782 – 790

Howe, H.F. & Lynn, L.C. (1988) Ecological Relationships of Plants and Animals. Oxford University Press, New York. Hurlbert, A.H., Hosoi, S.A., Temeles, E.J. & Ewald, P.W. (1996) Mobility of Impatiens capensis flowers: effect on pollen deposition and hummingbird foraging. Oecologia 105, 243 – 246. de Jong, T.J., Waser, N.M., Price, M.V. & Ring, R.M. (1992) Plant size, geitonogamy and seed set in Ipomopsis aggregata. Oecologia 89, 310 – 315. Kearns, C.A. & Inouye, D.W. (1993) Techniques for Pollination Biologists. University Press of Colorado, Niwot, CO. Kuijt, J. (1988) Revision of Tristerix (Loranthaceae). Systematic Botany Monographs 19, 1– 61. Lee, T.D. (1984) Patterns of fruit maturation: a gametophytic selection hypothesis. American Naturalist 123, 427– 443. Lee, T.D. (1988) Patterns of fruit and seed production. Plant Reproductive Strategies (eds J. Lovett Doust & L. Lovett Doust), pp. 177– 202. Oxford University Press, New York. Lloyd, D.G. & Barrett, S.C.H. (eds) (1996) Discovery of the secret of nature in the structure and fertilization of flowers. Floral Biology, Studies on Floral Evolution in AnimalPollinated Plants, pp. 3 – 43. Chapman & Hall, London. Marshall, D.L. & Folsom, M.W. (1991) Mate choice in plants: an anatomical to population perspective. Annual Review of Ecology and Systematics 22, 37– 63. Martin, F.M. (1959) Staining and observing pollen tubes by means of fluorescence. Stain Technology 34, 436 – 437. Mead, R. (1988) The Design of Experiments. Cambridge University Press, Cambridge, UK. Meléndez-Ackerman, E.J. (1997) Patterns of color and nectar variation across an Ipomopsis (Polemoniaceae) hybrid zone. American Journal of Botany 84, 41– 47. Meléndez-Ackerman, E.J., Campbell, D.R. & Waser, N.M. (1997) Hummingbird behavior and mechanisms of selection on flower color in Ipomopsis. Ecology 78, 2532 – 2541. Motten, A.F. & Antonovics, J. (1992) Determinants of outcrossing rate in a predominantly self-fertilizing weed, Datura stramonium (Solanaceae). American Journal of Botany 79, 419 – 427. Mulcahy, D.L. (1979) The rise of angiosperms: a genecological factor. Science 206, 20 – 23. Murcia, C. (1990) Effect of floral morphology and temperature on pollen receipt and removal in Ipomoea trichocarpa. Ecology 71, 1098 –1109. Murcia, C. & Feinsinger, P. (1996) Interspecific pollen loss by hummingbirds visiting flower mixtures: effects of floral architecture. Ecology 77, 550 – 560. van der Pijl, L. (1961) Ecological aspects of flower evolution. II. Zoophilous flower classes. Evolution 15, 403 – 416. Pleasants, J.M. & Chaplin, S.J. (1983) Nectar production rates of Asclepias quadrifolia: causes and consequences of individual variation. Oecologia 59, 232 – 238. Podolsky, R.D. (1992) Strange floral attractors: pollinator attraction and the evolution of plant sexual systems. Science 258, 791–793.

Podolsky, R.D. (1993) Evolution of flower dimorphism: how effective is pollen dispersal by ‘male’ flowers? Ecology 74, 2255 – 2260. Proctor, M., Yeo, P. & Lack, A. (1996) The Natural History of Pollination. HarperCollins, London. Rathcke, B.J. (1992) Nectar distributions, pollinator behavior, and plant reproductive success. Effects of Resource Distribution on Animal–Plant Interactions (eds M.D. Hunter, T. Ohgushi & P.W. Price), pp. 113–138. Academic Press, New York. Ruffini, A.E. (1992) Interacciones y coevolución entre el picaflor rubí (Sephanoides sephaniodes) y las plantas que poliniza en el bosque de Nothofagus. Licenciatura thesis, Universidad Nacional del Comahue, Bariloche, Argentina. SAS (1988) SAS/STAT User’s Guide, Release 6.03. SAS Institute, Cary, NC. Sazima, I., Buzato, S. & Sazima, M. (1996) An assemblage of hummingbird-pollinated flowers in a montane forest in southeastern Brazil. Botanica Acta 109, 149–160. Schoen, D.J. & Stewart, S.C. (1986) Variation in male reproductive investment and male reproductive success in white spruce. Evolution 40, 1109 –1120. Smith-Ramírez, C. (1993) Los picaflores y su recurso floral en el bosque templado de la isla de Chiloé, Chile. Revista Chilena de Historia Natural 66, 65–73. Sokal, R.R. & Rohlf, F.J. (1981) Biometry 2nd edn. Freeman, San Francisco, CA. Stanton, M.L. & Galen, C. (1989) Consequences of flower heliotropism for reproduction in an alpine buttercup (Ranunculus adoneus). Oecologia 78, 477–485. Stanton, M.L., Snow, A.A. & Handel, S.N. (1986) Floral evolution: attractiveness to pollinators increases male fitness. Science 232, 1625 –1627. Stiles, F.G. (1981) Geographical aspects of bird–flower coevolution, with a particular reference to Central America. Annals of the Missouri Botanical Garden 68, 323–351. Straw, R.M. (1956) Adaptive morphology of the Penstemon flower. Phytomorphology 6, 112–119. Sutherland, G.D. & Gass, C.L. (1995) Learning and remembering of spatial patterns by hummingbirds. Animal Behavior 50, 1273 –1286. Temeles, E.J. (1996) A new dimension to hummingbird–flower relationships. Oecologia 105, 517–523. Vaknin, Y., Tov, Y.Y. & Eisikowitch, D. (1996) Flowering seasonality and flower characteristics of Loranthus acaciae Zucc. (Loranthaceae): implications for advertisement and bird-pollination. Plant Sexual Reproduction 9, 279– 285. Waser, N.M. (1983) The adaptive nature of floral traits: ideas and evidence. Pollination Biology (ed. L.A. Real), pp. 241– 285. Academic Press, New York. Received 22 January 2001; revised 6 July 2001; accepted 10 July 2001