EffEct of nEctar sEcrEtion ratE on pollination succEss of Passiflora coccinea ( passifloracEaE ) in thE cEntral amazon fischEr

the pollination of Passiflora coccinea by the hummingbird Phaethornis superciliosus was studied in central amazon, Brazil. We hypothesized that a greater nectar secretion rate (nsr) increases the pollination success of single flowers through Ph. superciliosus visiting behavior. for control flowers, nsr was an increasing function of flower base diameter (fBD). the total number of Ph. superciliosus probes per flower was an increasing function of fBD. additionally, deposition of pollen on stigmas increased with the cumulative number of Ph. superciliosus probes. our results show that larger P. coccinea flowers secrete nectar at higher rates, are probed more times during each hummingbird visit and are more successful at pollination. this seems to be the first non-manipulative study describing such an effect of nsr on the pollination of single flowers in nature.


IntroductIon
flower availability and distribution have often been found to affect pollinator behavior, but how the nectar distribution among single flowers affects the visiting patterns of pollen vectors has received little attention (mitchell & Waser, 1992;rathcke, 1992).flower nectar distribution can determine pollinator behavior with respect to visitation frequency to flowering plants, to the number of flowers probed per visit, and to the number of probes per flower (fischer, 1992;Burd, 1995;conner & rush, 1996).Visits to single flowers can be an important component of pollinator foraging behavior that affects plant pollination.Based on experimental studies, the rate of nectar secretion by a given flower can affect the number and duration of pollinator probes, and ultimately its pollination success (Gill, 1988;mitchell & Waser, 1992;rathcke, 1992).one expects that flower traits have been selected in order to maximize pollen removal and deposition and to avoid self-pollination.however, this produces a dilemma for plants (Klinkhamer & Jong, 1993): how to be highly attractive to achieve effective pollination (female function) and at the same time induce visitors to travel to conspecific plants, so as to reduce geitonogamy and increase pollen exportation (male function).in addition, the rewards offered by a plant may involve a balance between the costs in producing them and the benefits gained from animal services (heinrich, 1975).
hummingbirds are nectar-dependent vertebrates that visit flowers to maintain a daily energy balance.several hermit hummingbirds (phaethorninae) are long-distance, high-reward trapliners (feinsinger, 1983).they are often the exclusive pollinators of particular plant species (e.g., stiles & freeman, 1993; araujo et al., 1994; 2004). in contrast to most non-hermit hummingbirds, the hermits commonly visit scattered flowering plants that only open a few flowers per day (feinsinger, 1983; araujo et al., 1994; 2004). in order to increase the rate of net energy intake, hermits tend to visit flowers at regular time intervals so that the amount of accumulated nectar is maximized relative to the energy spent in moving between plants (stiles & Wolf, 1979;Gill, 1988).
the genera Passiflora (passifloraceae) includes vine species pollinated by bees, hum-mingbirds, bats and wasps (e.g., sazima & sazima, 1978;Koschnitzke & sazima, 1997;Varassin et al., 2001;fischer et al., 2004).the hummingbirdpollinated P. speciosa and P. vitifolia present red and erect flowers exposed from above canopies, with approximately 20% nectar sugar concentration (Endress, 1994;proctor et al., 1996;longo, 2002).most floral features of P. coccinea are similar to P. speciosa and P. vitifolia, but its pollination was apparently not described.in addition, to our knowledge, the effect of nectar secretion rate (nsr) on the pollination success of single flowers has never been investigated under natural field conditions, by comparing natural variations among individual flowers.here we describe the pollination system of P. coccinea and ask if natural variation of nsr among non-manipulated flowers affects their pollination success by the long-tailed hermit Phaethornis superciliosus (trochilidae) visiting them.We hypothesized that the variation in nsr affects the number of hummingbird probes to individual flowers, and that the number of pollinator probes per flower affects the quantity of pollen deposited on the stigma surface.We previously determined that the diameter of the floral cup could be an index for nsr of P. coccinea flowers and measurements of nsr from flowers being visited in the field were unnecessary.

Study site and species
fieldwork was carried out on forest edges, between km 60-62 of the road manaus-caracaraí (Br 174), central amazon, Brazil (2° 35' s and 60° 2' W). the road crosses 30-40 m-tall undisturbed "terra firme" forest, bordered by growth vegetation.the studied 2-km road section contained twenty large flowering plants of P. coccinea.the species flowers from July to february in the central amazon (personal observations).plant voucher material was deposited at the unicamp herbarium (Euc).the hummingbird Phaethornis superciliosus is the largest hermit species at the study site and the only hummingbird observed to visit P. coccinea; it forages for nectar on scattered understory plants and on those along forest edges.at the study site, twenty ant species visited the extra floral nectaries (Efn) on P. coccinea floral bracts.Camponotus spp.(formicinae) and Crematogaster spp.(myrmicinae) were the most abundant and Ectatomma spp.(ponerinae) and Pseudomyrmexspp.(pseudomyrmecinae) the most aggressive.these ant species repelled visits of nymphalid and hesperiid butterflies, which acted as nectar thieves in flowers not patrolled by the ants; whereas hummingbirds were not disturbed by ants as they hovered during visits (leal et al., 2006).

Pollination biology
to determine plant pollination and the local pollinator species, flowering P. coccinea individuals were observed daily from august 26 to september 3, 1993.We counted the number of open flowers per plant and determined the flower lifetime.the time of nectar and pollen availability, receptivity of stigmas (cf., zeisler, 1938), and animal visits were recorded over the lifetime of flowers.nectar volume and nectar sugar concentration were measured unsystematically for flowers in several individuals at different hours and days with a syringe accurate to 1 µl and a pocket refractometer accurate to 1%. measurements of floral cups and androgynophores were done with a caliper accurate to 0.1 mm.activity and behavior of flower visitors were registered by direct observations and photographs.for focal plant individuals with different numbers of open flowers, the time and duration of visiting bouts and the number of hummingbird probes per flower were recorded.a visiting bout was defined as one hummingbird visiting any number of flowers before leaving the focal plant (fischer, 1992); and a probe was defined as each insertion of the bird's bill into the flower nectar chamber.since the number of open flowers on a plant might affect the frequency of visits, we tested if the mean time interval between Ph. superciliosus visits differed among plants with different numbers of open flowers.

Nectar secretion as a function of flower base diameter
to answer whether the nectar secretion rate (nsr) affects the number of hummingbird probes, we used the previously established relationship between nsr and the size of flower cups.the flower cup of P. coccinea has a circular base, so we used the easily measured external flower base diameter (fBD) as the independent variable.to test if nsr varies as a function of fBD, we selected five accessible flowers on different plant individuals that were open on the same day, so we standardized climatic conditions among flowers being measured.these five flowers were bagged before opening and the nectar volumes measured over three time intervals.flowers emptied soon after opening (0530 h) were then revisited three times (0630, 0750, and 0920 h), for measurements of accumulated nectar volume.Each time, we visited the five flowers in the same sequence to maintain similar time intervals among them.the duration of the first time interval was 80-85 min (81 ± 2.2 min), the second 105-110 min (107 ± 2.7 min) and the third 130-145 min (135 ± 7.1 min).the nsr per flower was calculated as the total volume of nectar secreted divided by the total time of nectar accumulation.the relation between nsr and fBD was evaluated by linear regression.

Nectar secretion effect on pollination of single flowers
to determine if nsr affects the number of pollinator probes, on 2 nd september eight focal flowers on eight different plant individuals were simultaneously observed by two people during their entire lifetime.these flowers were guarded by Efn-visiting ants, and visits from nectar thieves did not occur.Without disturbing the hummingbirds, the number of probes per flower was recorded for each flower's lifetime.soon after Ph. superciliosus finished each visiting bout, the stigmas were inspected and the proportion of their surfaces covered with pollen was recorded.the large (diameter = 4.1 ± 0.2 mm, n = 11) whitish stigma surface of P. coccinea allowed us to identify the yellow pollen area promptly.the proportion of pollen cover on each stigma was scored as a percentage of its surface and used as an index of pollination success (cf., heithaus et al., 1982).since a flower has three stigmas, we considered as 100% of stigma surface the sum of their areas.at the end of the flower anthesis, the stigmas of each of the eight flowers were collected to inspect pollen grains in the laboratory.regression analysis of the number of probes received per flower as a function of the fBD was performed.the spearman rank-correlation analysis was used to test for correlation between the cumulative proportion of pollen cover on stigmas and the cumulative number of hummingbird probes per flower.all means cited in the text are followed by ± standard deviations.

Pollination biology
the odorless, scarlet flowers of P. coccinea opened before sunrise (0500-0530 h) and lasted for less than a morning.the petals and bracts began to close at 1000 h, and all flowers were closed by 1130-1200 h.flowers were erect to slightly inclined, with radial symmetry (fig.1). the androgynophore length was 47.5 ± 3.8 mm (n = 12).flowering individuals opened 0-4 flowers per day (1.9 ± 1.5, n = 41).the anthers already presented pollen in pre-anthesis buds.the amount of pollen available in the anthers was higher at 0530 h and decreased with hummingbird visitation, being almost absent in most of the regularly visited flowers after 0900-0930 h.inspected stigmas (n = 21) were receptive after 0600 h and remained so at least until 1200 h.flowers are protandrous, as the anthers were ready to contact the pollinator's head 30-90 min before the stigmas were (for drawings of similar floral phases, see fig.1e, f in Buzato & franco, 1992).flowers began to secrete nectar at 0530-0600 h and stopped by 1030-1100 h.among flowers sampled unsystematically at different hours and days, the nectar volume was positively correlated with nectar sugar concentration (r = 0.80, p < 0.01, n = 22).among these flowers, the mean nectar volume and the standard deviation were greatest between 0830-0940 h, whereas the mean nectar sugar concentration decreased continuously throughout the morning (fig.2).
Phaethornis superciliosus was the only hummingbird recorded visiting P. coccinea flowers, although other hummingbird species occurred at the study site.its visiting activity started between 0530 and 0600 h and finished when the flowers stopped nectar secretion.individuals of Ph. superciliosus drank almost all the nectar from the flowers before leaving a plant; we found no nectar in the flowers just after visiting bouts.During a visiting bout, a hummingbird commonly visited all flowers of a plant and probed each flower more than once, by inserting the bill in different parts of the nectar chamber.the number of hummingbird probes per flower per visiting bout was 3.1 ± 1.5 (n = 36), and the cumulative number of probes per flower at the end of its lifetime was 16.1 ± 5.3 (n = 8).Daily, one or more Ph.superciliosus (we were unable to determine how many birds visited the plants) made 9-11 visiting bouts per flowering plant, with intervals averaging 18.3 ± 8.3 min (n = 13) between them.one visiting bout lasted 5-20 sec.Encounters between hummingbirds were not registered.
the sexual flower parts were touched by the hummingbird's crown in almost all probes (fig.1).however, in the earliest visits the flowers were still in the male phase, and Ph.superciliosus did not contact the stigmas.the time interval between visits was not different (f 2,19 = 1.84, p = 0.19) among plants with one, three or four simultaneously open flowers.nonetheless, plants with one open flower were visited with intervals (24.2 ± 9.3 min, n = 5) longer than those with three or four open flowers (respectively, 15.0 ± 7.9 min, n = 9; and 16.8 ± 9.5 min, n = 8).

Nectar secretion effect on pollination by hummingbirds
for the five bagged flowers monitored during one morning, the nectar secretion rate (µl.min -1 ) varied due to the flower base diameter (mm) (nsr = -4.24+ 0.35 fBD; p < 0.03; r 2 = 0.84).the mean nsr was 0.9 ± 0.13 µl.min -1 and the mean fBD was 14.6 ± 0.34 mm (n = 5).among these five flowers, both the nectar secretion rate and the nectar sugar concentration decreased throughout the morning (table 1). the number of Ph. superciliosus probes per flower of P. coccinea was greater in flowers with larger base diameters (Y = -52.4+ 4.4 X; r 2 = 0.81; p < 0.01; n = 8) (fig.3a). in addition, the proportion of the stigma surface covered with pollen was positively correlated with the cumulative number of hummingbird probes (spearman coefficient = 0.94; p < 0.01; n = 27) (fig.3b).cumulative pollen on stigmas increased slowly with the first probes.these probes occurred when most flowers were still in the male phase and hummingbirds rarely contacted stigmas at this time.following this relatively brief male phase, the  flowers (Y = -52.4+ 4.38 X; r 2 = 0.81, p < 0.01, n = 8); and b) cumulative proportion of the stigmatic surface (%) covered with pollen as a function of the cumulative number of probes by hummingbirds (spearman's coefficient = 0.94; p < 0.01).the sigmoid curve was the best-fitted equation (Y = 56/{1 + e -([X -10.5]/2.7)}).
proportion of stigmas covered with pollen quickly increased with additional Ph.superciliosus probes, until reaching a plateau after a flower received approximately 15 probes.later probes may not increase the pollen area because they tend to cover previous ones and because available pollen decreases as the morning advances.in the laboratory we found only P. coccinea pollen grains deposited on the stigmas of the eight flowers monitored for the number of Ph. superciliosus probes.

dIscussIon
Passiflora coccinea flowers appear to be specialized for pollination by long-billed hummingbirds, such as the long-tailed hermit at the study site.the color, morphology, time of passion floWEr nEctar sEcrEtion anD pollination Braz.J. Biol., 66(2B): [747][748][749][750][751][752][753][754]2006 anthesis, nectar volume and sugar concentration of P. coccinea flowers are compatible with traits described for most hummingbird-pollinated species, but contrasts with passionflowers pollinated by different vectors (Endress, 1994;Koschnitzke & sazima, 1997;Buzato et al., 2000;fischer et al., 2004).Passiflora coccinea plants open few flowers per day and secrete sugar-rich nectar over a short period, as already found for specialized flowers regarding pollinators (e.g., feinsinger, 1983; ashman & schoen, 1994; araujo et al., 1994; 2004).the flowers of P. coccinea place pollen exactly on the hummingbird's crown, as do several other species exclusively pollinated by hermit hummingbirds (e.g., araujo et al., 2004;Varassin et al., 2001).the increased length of the androgynophore indicates that P. coccinea flowers are adapted for pollination by long-billed rather than by short-billed hummingbirds (sensu feinsinger; 1983).it seems likely that hummingbirds with bills shorter than that of Ph. superciliosus would be less effective for pollination since their heads would not touch anthers and stigmas at each visit. in the peruvian amazon, P. coccinea flowers are nectar-robbed by little hermit hummingbirds (J.m. olesen, personal communication).
a positive relation between nectar volume and nectar sugar concentration like that found in P. coccinea flowers has been previously described for the hummingbird-pollinated P. speciosa Gardn. in the southern pantanal, and for some bat-pollinated flowers in central amazon (fischer, 2000; longo & fischer, 2006).these results indicate that nectar sugar concentration increases while the secreted nectar accumulates, so visitors may find a higher sugar concentration if they return to flowers after total nectar replenishment.Passiflora coccinea flowers with larger base diameters secrete nectar faster.the variation in nectar secretion rate due to the flower base diameter might allow us to study the nectar intake by Ph. superciliosus.however, the nectar secretion rate may continuously decrease as the accumulated nectar volume increases (Galetto et al., 1994;Varassin et al., 2001), indicating that large flowers, if visited more often, might secrete even more nectar.in the five bagged flowers we measured the nectar in intervals longer than those between visiting bouts by Ph. superciliosus, thus the estimated function for nsr = f (fBD) may not accurately estimate the nectar intake by this hummingbird.
considering that Ph. superciliosus probed more times those P. coccinea flowers with a larger base diameter -and assuming the flower base diameter is an index of nectar secretion rate -it may be concluded that hummingbirds probe more times from flowers that secrete nectar at higher rates.the visiting behavior of Ph. superciliosus, therefore, seems sensitive to the variation of the nectar secretion rate by single P. coccinea flowers at the study site.these results for P. coccinea under natural conditions are consistent with experimental data for Ipomopsis (mitchell & Waser, 1992).
the number of flowers that open each day can also affect the visit frequency of pollinators to individual plants (e.g., trombulak, 1990; fischer, 1992; conner & rush, 1996).however, for P. coccinea we did not find an effect of the number of open flowers per plant on the time interval between Ph. superciliosus visits.the absence of such an effect supports the fact that the nectar secretion rate alone determines the variation in the number of Ph. superciliosus probes in P. coccinea flowers at the study site.
the proportion of the stigmatic surface covered with pollen increased asymptotically with the number of hummingbird probes, reinforcing the fact that pollination success of P. coccinea flowers is strongly affected by the Ph.superciliosus visiting behavior at the study site.Because the mean number of hummingbird probes per flower is higher than 15 (16.1 ± 5.3; n = 8), and considering that the stigmatic area covered with pollen reaches a plateau after approximately 15 probes, all the P. coccinea flowers studied by us appeared to reach high pollination success through the Ph.superciliosus visits.We highlight that the proportion of stigmatic area covered with pollen is only a rough estimator for pollination success (see other procedures in Dafni, 1992) and, additionally, we do not know the amount of pollen required to maximize a seed set. in Passiflora speciosa the number of hummingbird probes explained approximately 30% of the total variation in the number of seeds per fruit (longo & fischer, 2006).the present study allows one to speculate that, ultimately, the large flower size increases the reproductive success of P. coccinea and, therefore, the flower size could be a subject of selection through the Ph.superciliosus visiting behavior at the study site (see Burd, 1995;campbell et al., 1996).

Fig. 1 -Fig. 2 -
Fig. 1 -Phaethornis superciliosus visiting a bright red Passiflora coccinea aubl.flower in the central amazon, Brazil.note that the bill and the androgynophore lengths combine to determine the effective contact between the bird's head and the sexual flower parts.