Dark purple nectar as a foraging signal in a bird-pollinated Himalayan plant

Authors

  • Feng-Ping Zhang,

    1. Key Laboratory of Biodiversity and Biogeography, Kunming Institute of Botany, Chinese Academy of Sciences, Kunming 650204, China
    Search for more papers by this author
  • Xiang-Hai Cai,

    1. State Key Laboratory of Phytochemistry and Plant Resources in West China, Kunming Institute of Botany, Chinese Academy of Sciences, Kunming 650204, China
    Search for more papers by this author
  • Hong Wang,

    1. Key Laboratory of Biodiversity and Biogeography, Kunming Institute of Botany, Chinese Academy of Sciences, Kunming 650204, China
    Search for more papers by this author
  • Zong-Xin Ren,

    1. Plant Germplasm and Genomics Center, Germplasm Bank of Wild Species, Kunming Institute of Botany, Chinese Academy of Sciences, Kunming 650204, China
    Search for more papers by this author
  • Zachary Larson-Rabin,

    1. Plant Germplasm and Genomics Center, Germplasm Bank of Wild Species, Kunming Institute of Botany, Chinese Academy of Sciences, Kunming 650204, China
    Search for more papers by this author
  • De-Zhu Li

    1. Key Laboratory of Biodiversity and Biogeography, Kunming Institute of Botany, Chinese Academy of Sciences, Kunming 650204, China
    2. Plant Germplasm and Genomics Center, Germplasm Bank of Wild Species, Kunming Institute of Botany, Chinese Academy of Sciences, Kunming 650204, China
    Search for more papers by this author

Authors for correspondence:
De-Zhu Li
Tel: +86 871 5223503
Email: dzl@mail.kib.ac.cn
Hong Wang
Tel: +86 871 5223534
Email: wanghong@mail.kib.ac.cn

Summary

  • Some plants secrete coloured nectar to attract pollinators, but little is known about the chemical origins of nectar colouration and its ecological function. Leucosceptrum canum stands out as the only plant with coloured nectar recorded in the Himalayas. Here, we focused on the compound associated with the dark colour of the nectar, as well as its secretion dynamics during the flowering season and its relationship to pollinators.
  • Fresh nectar was analysed by semi-preparative reversed-phase high-performance liquid chromatography (HPLC), LC-MS and HRESIMS (high resolution electronspray ionization mass spectroscopy) to determine which compound causes the nectar colouration. Behavioural experiments were conducted with birds and honeybees to elucidate the effect of the nectar colour and volume on pollinators.
  • We identified a purple anthocyanidin, 5-hydroxyflavylium, as a natural nectar product for the first time. Two short-billed birds were found to pollinate this plant, which employs two nectar-based mechanisms to direct bird pollinators to reproductively active flowers, controlling nectar palatability and presenting a foraging signal for birds by altering nectar volume and colour in a developmental stage-specific manner.
  • 5-Hydroxyflavylium was found to be the cause of the nectar colouration, the function of which is to act as a foraging signal to increase pollination efficiency through nectar visibility and palatability.

Introduction

The co-evolution between visual signals produced by plants and the behavioural responses of animal mutualists is one of the most intriguing subjects in the biological sciences. Chemical compounds in floral nectar can be diverse and abundant (Baker, 1977; Bernardello et al., 1994; Adler, 2000), and some studies have suggested that they attract nectarivores (Janzen, 1977; Olesen et al., 1998; Adler, 2000; Johnson et al., 2006). Certain compounds have been found to influence nectar colouration (Mione & Anderson, 1996; Olesen et al., 1998; Weller et al., 1998) and taste (Hagler & Buchmann, 1993; Adler, 2000; Gardener & Gillman, 2002). Several flowering plants possess coloured nectar, and the first recorded observation of this rare floral trait was by the Greek poet Homer (Hansen et al., 2007), but it has been somewhat under-researched by modern science.

Thus, little is yet known about the chemical origins of nectar colouration and its ecological functions. Olesen et al. (1998) discovered the first compound responsible for nectar colouration by studying a bird-pollinated plant in Mauritius, Nescodon mauritianus (Campanulaceae); this species’ scarlet-red nectar colouration was found to be caused by a red aurone. It was postulated that the coloured nectar may serve as an honest signal to bird pollinators, although the function of this colouration remained a mystery because of a lack of experimental data. Recent ecological experiments revealed that three Mauritian plants with coloured nectar were pollinated by the Phelsuma geckos (Phelsuma ornata and Phelsuma cepediana), which used the nectar colour as a visual signal (Hansen et al., 2006). Johnson et al. (2006) reported that a South African Aloe (Aloe vryheidensis) uses its dark nectar as a floral filter, attracting bird pollinators visually and repelling nectar robbers with its bitter taste.

Most of the known coloured-nectar plants occur in tropical and subtropical regions of the Southern Hemisphere (Hansen et al., 2007). Only one flowering plant species with coloured nectar, Leucosceptrum canum (Labiatae), has been reported in the Himalayas and southwestern China (Hansen et al., 2007). It has been assumed that this plant is ornithophilous (Cowan & Cowan, 1929), but neither the chemical origin nor the functional significance of the nectar’s colour has been characterized.

Here, we studied the chemical compound associated with the dark purple nectar of Lcanum, as well as its secretion dynamics during the flowering season and its relationship to pollinators. The aims of this study were to determine which compound causes the nectar colouration, to elucidate the ecological function of the coloured nectar in reproduction, and to evaluate the impact of dynamical variance of coloured nectar on pollinators.

Materials and Methods

Study species and sites

Leucosceptrum canum Smith is a shrub or small tree that grows to up to 10 m in height at altitudes of between 1000 and 2600 m in the Sino-Himalayas, and currently represents the only Labiatae species to produce coloured nectar. The inflorescences are composed of brushy panicles, densely packed with small, cream-white flowers. It flowers from mid November to early March. This plant was studied from December 2008 to March 2011. Fieldwork was conducted on Bagua Mountain, Baoshan, western Yunnan, China (geographical coordinates 25°02′29′′N, 99°51′27′′ E). On the basis of field observations, behavioural experiments were designed and carried out at the Kunming Botanical Garden, Kunming, Yunnan, China (25°08′42′′N, 102°44′31′′E).

Nectar and flower traits

Nectar volume and sugar concentrations were determined using micro-capillaries (5, 10, 20 and 50 μl) and a pocket refractometer (Eclipse 45-81; Bellingham and Stanley Ltd, Tunbridge Wells, Kent, UK). Ninety mature flowers (stage 3) of nine inflorescences from three plants were randomly selected for measurements of floral traits, including the number of flowers per inflorescence, style length, stamen length, and corolla tube depth.

Fifty floral buds on three plants were observed for 10 d to record the timing of flower developmental events, including stamen and style elongation, anther dehiscence and the presence of nectar. Following Nicolson & Nepi (2005), flower development was divided into four stages: anthesis (stage 1, premature); emergence of style and stamens (stage 2, premature); opening of anthers, dehiscence of pollen, and initiation of stigma receptivity (stage 3, mature); wilting of corolla tube and stamens (stage 4, senescent).

To assess the effect of the age of the flower on nectar secretion patterns, flower buds were labelled and covered with fine-mesh nylon bags (250 × 200 mm). Nectar volume and sugar concentrations in 50 flowers of each development stage on each of three plants were measured. Nectar volumes and sugar concentrations of 30 uncovered stage 3 flowers from each of three plants were also measured every 2 h over a 24-h period, with each flower sampled only once.

Chemical experiments

Fresh nectar (1 ml) was separated by semi-preparative reversed-phase C18 high-performance liquid chromatography (HPLC) on a Sunfire column (150 × 10 mm) with gradient flow from 25 to 30% aqueous methanol to detect its chemical components. LC-MS analysis was performed using Bruker Compass Data Analysis 4.0 (Bruker company, Bremen, Germany) (250 × 4.6 mm, gradient flow from 25 to 35% aq. MeOH). Compound isolation was performed on Waters 600 and 2695 pumps employing a Waters 2996 photodiode array detector (Waters Company, MA, USA). HRESIMS (high resolution electronspray ionization mass spectroscopy) was obtained on an API Qstar Pulsar I spectrometer (Applied Biosystems Ltd., Warrington, UK).

Pollinator observations and pollination experiments

In open pollination, observations were conducted between 09:00 and 17:00 h for 10 successive days on Bagua Mountain. All bird and insect visitors were identified, and their behaviours were recorded and photographed. The presence of pollen grains was confirmed on birds captured in mist nets, by collecting pollen samples from their faces and comparing these to a reference pollen collection (Johnson et al., 2006). The lengths of the bills of the captured birds were measured.

Five pollination experiments were conducted on separate groups of three plants each in the study site in January 2010: (1) natural pollination: flowers were not manipulated (= 450); (2) bird-exclusion pollination: floral buds were bagged with plastic mesh (20-mm-diameter apertures) to exclude birds without preventing visitation by honeybees (= 450); (3) autonomous self-pollination: floral buds were bagged with solid plastic throughout their flowering period (= 450); (4) hand self-pollination: bagged mature flowers (stage 3) were hand-pollinated with pollen from the same flower (= 407); (5) hand cross-pollination: bagged flowers were emasculated before anthesis and mature flowers (stage 3) were pollinated with stage 3 pollen from plants growing 10 m away (= 324). Mature fruits and seeds were collected and counted.

Behavioural experiments

To elucidate pollinator behaviour, controlled experiments were conducted based on field observations. The two major bird pollinators were the blue-winged minla (Minla cyanouroptera) and the oriental white-eye (Zosterops palpebrosa). Before beginning the behavioural experiments, birds were maintained in large outdoor aviaries (3 m2). Responses of birds to the nectar were determined in experiments conducted individually in a smaller outdoor aviary (60 × 60 × 60 cm). Naïve individuals were used for each behavioural experiment.

To ascertain whether these species would prefer the coloured over clear nectar, birds (blue-winged minla, = 20; oriental white-eye, = 20) were offered a choice between two model flowers (base sections of 0.75-ml Eppendorf tubes cut in half and painted white), which were identical in all respects except that the base of one was painted black (Johnson et al., 2006). The model flowers were filled with 50 μl of a 15% hexose solution (glucose and fructose in a 1 : 1 mixture) and were placed 5 cm apart above a suitable perch using double-sided adhesive tape. The first model flower to be probed by a bird was recorded in each trial.

To establish whether pollinators would adjust their visiting behaviour according to the presence, absence, or volume of nectar (blue-winged minla, = 20; oriental white-eye, = 20), white-painted model flowers were filled with nectar of stage 3 L. canum flowers in volumes ranging from to 0 to 60 μl (0 vs 5 μl, 5 vs 15 μl, 15 vs 30 μl, 30 vs 45 μl, and 45 vs 60 μl) and were placed 5 cm apart above a suitable perch using double-sided tape. The first model flower to be selected by a bird was recorded in each?trial.

To define the palatability of Lcanum nectar to birds (blue-winged minla, = 20; oriental white-eye, = 20), feeding trials involving choices among 50 μl nectar (taken from stage 3 flowers) and solutions of 15% hexose or 15% sucrose were established. The sugar solutions and nectar were offered to birds simultaneously in three model flowers. Each trial was terminated after a bird had probed all three of the model flowers. The volumes of sugar solutions and nectar remaining at the end of the experiment were determined using a calibrated micropipette, and the model flowers were replenished for the subsequent trial.

To determine the palatability of Lcanum nectar to honeybees, honeybees (= 35) were captured and kept in individual containers (25 × 25 × 25 cm). They were then offered nectar (taken from stage 3 flowers) and 15% sugar solutions in the form of 5-μl droplets on a 10-cm-diameter white plastic disc. In one set of trials, honeybees were offered a choice between two 5-μl droplets of nectar and two 5-μl droplets of hexose solution, while in the second set of trials the hexose solution was replaced with the sucrose solution. Each trial was terminated after an individual bee had probed all four droplets on the disk. The remaining volume of each droplet was measured using a calibrated micropipette.

To compare the palatability of nectar from the different flower development stages to birds, feeding trials involving choices among 50-μl nectar samples from stages 1/2, stage 3, and stage 4 were carried out. Nectar samples from stage 1 and stage 2 were combined because they did not differ in sugar concentration. The three samples of nectar were offered to birds simultaneously in three white model flowers. Each trial ended when a bird had probed all three of the model flowers, and the remaining nectar volume of each flower was determined. After each trial the sugar/nectar solution was replaced and the position of each model flowers was randomized. Meanwhile, the bird’s behaviour was observed and recorded.

Statistical analysis

Kruskal–Wallis tests were used to determine the statistical significance of differences in the volume and sugar concentration of nectar from the four floral stages. Paired t-tests were used to analyse the preference trials. Kruskal–Wallis and post hoc (LSD) tests were also used to compare the palatability to birds of the hexose and sucrose solutions and the nectar from the different floral stages. All analyses were performed using spss 16 (SPSS Company, Chicago, Illinois, USA), with measured variables presented as mean ± SE.

Results

Flower and nectar traits

Leucosceptrum cannum flowers at the bottom of the brushy panicle inflorescences opened first, and maturation passed in a wave across the inflorescences in c. 1 wk, resulting in clumps of mature flowers at different locations during different periods. Four developmental events were recognized, starting with the opening of the corolla tube (stage 1). Twenty-four hours later, the style and stamens emerged but the anthers remained closed (stage 2). After another 24 h, the style and stamens elongated, anthers opened, and pollen dehisced (stage 3). The corolla and stamens began to wilt 96 h after anthesis (stage 4), although the style remained turgid until 120 h. At 144 h the floral tube, stamens, and style were completely senesced.

Nectar of L. canum was initially secreted as a light purple liquid during premature stages (i.e. stages 1 and 2), but it became dark purple as flowers matured (stage 3) (Fig. 1a). There was a consistent increase in both nectar volume and sugar concentrations as flowers progressed from stage 1 to stage 3. The average nectar volume per flower increased from 0.75 μl in stage 1 to 12.34 μl in stage 3, while in stage 4 nectar volume declined to 5.41 μl (χ2 = 479.90; < 0.001) although its sugar concentration increased (χ2 = 456.91; < 0.001) (Fig. 2). Periodic sampling from stage 3 flowers throughout the day and night showed that there were no significant differences in nectar volume (χ2 = 2.61; = 1.00) or sugar concentration (χ2 = 4.23; = 0.98), even though temperature and relative humidity varied diurnally (Fig. 3). Nor was a change in nectar colour detected between day and night. Each inflorescence bore 458.27 ± 12.60 flowers (= 75) with an average floral-tube length of 9.56 ± 0.05 mm (n = 90), and an average style length of 24.37 ± 0.15 mm (= 90) at stage 3.

Figure 1.

Leucosceptrum canum flowers and pollinator. (a) The dark purple nectar (arrow); (b) the structure of the anthocyanidin; (c) a blue-winged minla feeding from flowers of L. canum.

Figure 2.

Variation in the volume and sugar concentration of Leucosceptrum canum nectar at different floral ages (stages 1, 2, 3 and 4) on three individuals. (a) Nectar volume; (b) nectar sugar concentration. All flowers were bagged. Both nectar volume and sugar concentration increased during flower development through stage 3, but in stage 4 nectar volume declined while sugar concentration increased. Different lowercase letters indicate significant statistical differences, while the same letters indicate no significant difference. Error bars indicate ± SE.

Figure 3.

Nectar volumes and sugar concentrations in unprotected flowers of?Leucosceptrum canum were measured throughout the day. (a) Nectar volume (= 390); (b) nectar sugar concentration (= 390); (c) temperature (closed circles) and relative humidity (open circles). Error bars indicate?± SE.

Chemical experiments

Using a range of analysis tools, we isolated a purple compound from the nectar. Its UV spectrum showed the characteristic absorption bands of anthocyanidin at 226, 267, and 507 nm. This new isolate was found to possess a molecular formula of C15H10O2 as shown by HRESIMS at m/z 222.0681 [M]+ (calculated for C15H10O2, 222.0680), inconsistent with its tautomer (quinoid form). Thus it was identified as 5-hydroxyflavylium (Fig. 1b). During separation and isolation of the nectar components, we found that this compound was the cause of the nectar colouration. Because this compound is unstable, NMR spectroscopic data were not obtained.

Pollinator observations and pollination experiments

Flowers of L. canum were visited by 11 bird species and one honeybee species during the field observation period, and the nectar was fed upon by both birds and honeybees. The most common visitors among these birds were the blue-winged minla (Fig. 1c) and the oriental white-eye. The faces of these two species were observed carrying visible quantities of Lcanum pollen grains. The average lengths of the protruding sections of the styles and stamens of stage 3 flowers (14.81 ± 0.15 mm, = 90, and 16.23 ± 0.13 mm, = 90, respectively) were longer than the average lengths of the birds’ bills (blue-winged minla: 12.68 ± 0.14 mm, = 20; oriental white-eye: 10.03 ± 0.12 mm, = 20), and we visually confirmed that the faces of these birds made effective contact with the anthers and stigmas during nectar consumption. Both of the bird species showed a preference for stage 3 flowers around the middle of the inflorescences.

We concluded that the blue-winged minla and oriental white-eye were the main pollinators of this plant species. These birds are distributed across the Sino-Himalayas, with the range of the blue-winged minla extending from south-central China to mainland Southeast Asia, and the range of the oriental white-eye extending to mainland Southeast Asia and overlapping with that of L. canum in the Himalayas and southwestern China.

This plant is self-compatible but dependent on pollinators to increase reproductive output. The fruit set per flower on inflorescences from which birds were excluded was significantly lower than for open-pollinated inflorescences (32.22% vs 69.78%; < 0.001, paired t-test). Fruit set of self-pollination (71.01%) was similar to that of cross-pollination (73.41%) (Table 1).

Table 1.   Fruit set and seed production in pollination experiments on Leucosceptrum canum
TreatmentnFruit set (mean ± SE)Seed production (mean ± SE)
  1. Different letters indicate significant statistical differences, while the same letters indicate no significant difference.

Self-pollination40771.01 ± 1.18a1.95 ± 0.03a
Cross-pollination32473.41 ± 0.94ab1.96 ± 0.03a
Bird-exclusion pollination45032.22 ± 1.06b1.45 ± 0.03b
Bagged control45030.00 ± 1.45b1.43 ± 0.05b
Open-pollination45069.78 ± 0.83ac1.97 ± 0.04a

Behavioural experiments

Behavioural experiments showed that the blue-winged minla and the oriental white-eye foraged mainly for coloured rather than clear nectar in model flowers (Fig. 4). These two bird species preferred larger nectar volumes, up to 30 μl (blue-winged minla: 0 vs 5 μl: χ2 = 29.74; < 0.001; 5 vs 15 μl: χ2 = 29.35; < 0.001; 15 vs 30 μl: χ2 = 30.28; < 0.001; = 200; oriental white-eye: 0 vs 5 μl: χ2 = 30.01; < 0.001; 5 vs 15 μl: χ2 = 28.67; < 0.001; 15 vs 30 μl: χ2 = 25.42; < 0.001; = 200). Neither species differed markedly in its response to the nectar volumes above 30 μl (blue-winged minla: 30 vs 45 μl: χ= 2.19; = 0.139; 45 vs 60 μl: χ2 = 0.13; = 0.724; = 200; oriental white-eye: 30 vs 45 μl: χ2 = 1.88; = 0.17; 45 vs 60 μl: χ2 = 2.19; = 0.14; = 200) (Fig. 5). They adjusted their behaviour according to the nectar presence, absence, or volume difference before visiting flowers (that is, they decided which flower to visit before reaching the flowers).

Figure 4.

First choices between dark-centred and all-white model flowers. (a) Blue-winged minla; (b) oriental white-eye.

Figure 5.

First choices between volumes (dark grey bars, lower volume; light grey bars, higher volume) for pollinators. (a) Blue-winged minla; (b) oriental white-eye. Different lowercase letters indicate significant statistical differences, while the same letters indicate no significant difference.

Feeding trials showed that the birds consumed similar volumes of L. canum nectar and hexose and sucrose solutions (blue-winged minla: χ2 = 2.13; = 0.35; = 200; oriental white-eye: χ2 = 5.04; = 0.08; = 200). The birds preferred nectar of stage 3 flowers but still consumed nectar of floral stages 1 and 2 (premature) (blue-winged minla: χ2 = 35.06; < 0.001; = 18; oriental white-eye: χ2 = 36.82; < 0.001; = 17) (Fig. 6). However, the nectar was unpalatable during floral stages 1 and 2, and the birds shook their heads after probing the nectar of such flowers. Also, they wiped their bills rapidly after foraging the nectar of stage 4 flowers (senescent). Conversely, the birds showed no adverse reaction while foraging the nectar of stage 3 flowers.

Figure 6.

Preferences of Leucosceptrum canum nectar from stages 1/2, stage 3, and stage 4 flowers, when offered a choice. (a) Blue-winged minla (post hoc test (LSD): stages1/2 vs stage 3: < 0.001; stages 1/2 vs stage 4: < 0.001; stage 3 vs stage 4: = 0.81); (b) oriental white-eye (post hoc test (LSD): stages1/2 vs stage 3: < 0.001; stages 1/2 vs stage 4: < 0.001; stage 3 vs stage 4: = 0.78). Error bars indicate ± SE.

Discussion

Flower nectar is secreted by plants to attract and reward pollinators, and it plays a key role in the functional ecology of plants (Baker & Baker, 1983; Gonzalez-Teuber & Heil, 2009). In general, sugars and amino acids are regarded as the main attractive components of the nectar (Gonzalez-Teuber & Heil, 2009). Secondary compounds have been found to change the pigmentation (Mione & Anderson, 1996; Olesen et al., 1998; Weller et al., 1998) and taste (Hagler & Buchmann, 1993; Adler, 2000; Gardener & Gillman, 2002) of nectar.

A chemical study on the dry flowers of L. cannum identified a yellow amine pigment (Huang et al., 2004). However, this amine was not detected in the nectar, so the possibility that it had a role in nectar colouration could be excluded. Pu & Zhou (1989) isolated three chemical components from leaves of L. canum that are not related to nectar colouration. Choudhary et al. (2004) isolated a novel sesterterpene from the aerial part of the plant. Further chemical studies on the leaves of Lcanum identified four other sesterterpenoids, and antifeedant experiments showed that two of them were potential deterrents to beet armyworm and cotton bollworm (Luo et al., 2010, 2011). However, no chemical compounds responsible for the colouration of the nectar were isolated. At the beginning of our study, we concentrated our efforts on the isolation of natural compounds in floral stage 3, as this stage produced the most abundant nectar. However, the dominant constituents were sugars during stage 3, and it was difficult to isolate other compounds at this stage, although we searched extensively for phenolic compounds which may be oxidized into quinoids (Parveen et al., 2010). After a number of failed trials, we shifted focus to the nectar from stage 1 and stage 2 flowers, and thereby found the purple anthocyanidin 5-hydroxyflavylium. Although we found the compound to be unstable when isolated, the purple colour in whole nectar is stable. Probably as a result of the compound’s instability as well as the high sugar concentration in stage 3 nectar, it was difficult to detect 5-hydroxyflavylium in stage 3 flowers. Alternatively, the compound may have been oxidized to another compound in later floral stages. Nevertheless, as it was the only secondary compound isolated from L. canum nectar, we suggest that 5-hydroxyflavylium is the cause of the nectar colouration.

The dynamic production of nectar is a potential adaptive strategy in flowering plants (Nepi et al., 2001; Luyt & Johnson, 2002; Nicolson & Nepi, 2005; Human & Nicolson, 2008). Our results showed that the colour, volume, sugar concentration, and taste of L. canum nectar changed at different developmental stages. This colour change may be attributable to: oxidation of the 5-hydroxyflavylium; the increase in sugar concentration; or production of a yet unidentified compound (Nicolson & Nepi, 2005; Johnson et al., 2006). However, nectar colour, volume and sugar content did not vary between day and night, as has been observed in Aloe (Nicolson & Nepi, 2005).

As L. canum nectar is fed upon by many birds, it has been assumed that this plant is ornithophilous (Cowan & Cowan, 1929; Hansen et al., 2007). Our results indicate that Lcanum is self-compatible but bird pollinator-dependent for increasing pollination efficiency. The short-billed faces of the blue-winged minla and the oriental white-eye were observed carrying visible quantities of Lcanum pollen grains, and the birds’ faces made contact with both anthers and stigmas during foraging. Although the fire-tailed sunbird (Aethopyga ignicauda) was common at the study site, the longer, narrower bill of this species was unsuitable for pollination of Lcanum. From a biogeographical consideration of the overlapping distribution patterns of the blue-winged minla, the oriental white-eye, and Lcanum, we could surmise that the shapes of the birds’ bills are adaptive. The bird-exclusion experiments showed that honeybees do not contribute to seed production, as their small bodies do not make contact with the stigmas and anthers while feeding on nectar.

Coloured nectar, although rare among flowering plants, can be an important floral trait in the mutualism between plants and their animal pollinators, and it has been regarded as a visual signal for pollinators to enhance reproductive success (Olesen et al., 1998; Hansen et al., 2006, 2007; Johnson et al., 2006). Our behavioural experiments showed that the dark purple nectar of Lcanum acts as a foraging signal to bird pollinators. Another possible explanation for the presence of the colour, however, may be that 5-hydroxyflavium inhibits bacterial or fungal growth, but this possibility is not mutually exclusive with the honest-signal function of the nectar. Future studies should determine whether female reproductive success is enhanced by the presence of this coloured nectar, by replacing the nectar with a clear sugar solution at stage 3 (Hansen et al., 2007).

The discovery of the coloured nectar’s role as a foraging signal is an important contribution to the understanding of the evolution of coloured nectar. Floral visual attractiveness, and nectar and pollen rewards for pollinators, are common features across the Labiatae (Huck, 1992; Harley et al., 2004), but this study is the first time that the visual signal has been recorded as deriving from the nectar itself. L. canum’s pollinators can detect the visible reward and reorient their trajectories towards rewarding flowers to improve the reproductive output of the plant (de Jong et al., 1993). We demonstrated for the first time that the presence/absence of coloured nectar or differences in nectar volume can influence bird movements, also enhancing reproductive success. For instance, flowers showing higher volumes of coloured nectar would be more likely to be both mature (i.e. stage 3) and unvisited; being mature is necessary for the transfer of pollen to and from the pollinator (as this is the stage when the stamens dehisce and stigmas become receptive), and being unvisited would increase the chance for a higher number of a plant’s flowers to be visited at least once, thus increasing reproductive output.

Under circumstances in which the nectar volume is larger and the reward is greater, the nectar is also more visible, and hence the visual attraction at a distance is stronger (Hansen et al., 2007). The active regulation of nectar production by flowers has been suggested to be an adaptive trait to improve reproduction success (Nicolson, 1995; Biernaskie et al., 2002). The results of our bird behaviour trials confirmed the signal value of the presence or absence and the volume of the coloured nectar.

Interestingly, the apparently disagreeable taste of nectar in stage 1, stage 2, and stage 4 flowers represents another mechanism by which L. canum can influence the stage at which bird visitation occurs. This second mechanism corrects the behaviour of birds that happen to feed on flowers of the other, nonreproductively capable stages. A visitor to a stage 2 flower, for example, would dislike the reward and so may learn to avoid stage 2 flowers thereafter. Thus, L. canum has evolved two separate nectar-based strategies to elicit bird visitation only to its reproductively active flowers. Besides increasing reproductive success directly, together these two strategies also keep immature flowers from being damaged or depleted of nectar by precocious visitation.

It is puzzling that coloured nectar is not more prevalent across the angiosperms. Hansen et al. (2007) described several characteristics shared by many plants with coloured nectar, and L. canum shows most of these: for example, the coupling of signal and reward, the production of many flowers at different rates of maturation, and the increase in floral signal value over developmental time. Yet, many plants that also share these characteristics do not produce coloured nectar. Perhaps the factors that must align for efficient pollination (e.g. the timing of floral developmental progress, the rate of nectar constituent production, the presence of compatible birds and their ability to learn) are too numerous for these strategies to evolve frequently. Alternatively, perhaps nectar is simply a less reliable medium for displaying colour, as the medium itself can evaporate or be stolen, whereas showy perianth tissues are more durable. The flowers of L. canum have little perianth tissue with which to signal, and, while their white and beige colours provide excellent contrast with the dark nectar (Fig. 1a,c), in the absence of the nectar one would expect the flowers to be rather unattractive to birds. In this case, the plant does not need to expend resources on corolla enlargement or perianth pigment production, as the signalling is performed by the nectar, and thus the corolla is white and practically vestigial. However, for the coloured nectar to assume the floral signalling function, it must first arise, and the cause of the original production of 5-hydroxyflavylium in the nectar remains a mystery.

In this study, we have isolated a new nectar anthocyanidin that may be responsible for both the colour and the taste of L. canum nectar. We also showed that the flowers of L. canum employ two mechanisms with which to direct bird pollinators to reproductively active flowers: controlling nectar palatability, and altering nectar volume and colour to manipulate the floral display. Comparative studies of the pollination biology of species closely related to L. canum would be particularly useful in understanding the evolution of coloured nectar in this lineage.

Acknowledgements

We thank R. F. Lu, S. Zhang, H. L. Ai, J. Liu and M. Y. Zhang for their kind assistance in the field, G. Y. Yang, X. D. Luo and Q. S. Zhao for suggestions regarding the nectar chemical analysis, and three anonymous reviewers for comments that improved the manuscript. The study was supported by the National Basic Research Program of China (973 Program, 2007CB411603), Key Laboratory of Biodiversity and Biogeography, Kunming Institute of Botany (KIB), Chinese Academy of Sciences (CAS) (No. 52Y0264111L1); the Science Foundation of Yunnan Province, China (No. 312010CD105); the State Key Laboratory of Phytochemistry and Plant Resources in West China, KIB, CAS (No. 52P2010-KF05) and the ‘Western Light’ project of CAS (No. 292010312D11036). Z.L.-R. was supported by a Chinese Academy of Sciences Fellowship for Young International Scientists.

Ancillary