To minimize interspecific pollination, it has been suggested that pollen is placed on different parts of a pollinator's body corresponding to the conspecific location of pollen pickup by the stigma.
Although Pedicularis is regarded as a classic example of pollinator-mediated floral isolation, such reciprocal pollen placement has not been demonstrated experimentally. This leads us to question previous observations of pollen release in Pedicularis species.
Here, we show that pollen grains are released from the tip, rather than the basal opening, of the galea (the hoodlike upper lip of the corolla) in eight nectarless Pedicularis species, mimicking pollen release from poricidal anthers. We used safranin-stained pollen within anthers to track pollen placement in three Pedicularis species, and showed that pollen was deposited on numerous parts of the bumblebee's body. However, fluorescent powder placed on the stigmas to detect the contact location on the bumblebee's body was deposited mainly on the major position of pollen placement in each of the three species.
Such segregation of pollen placement and pickup between species sharing the same pollinator probably helps to reduce reproductive interference, but the positions of pollen placement and stigma contact on the bumblebee's body were not as precise as previously thought.
Ecological and evolutionary processes have led to assemblages of species that are likely to coexist in ways that minimize reproductive interference (Coyne & Orr, 2004). When multiple species share the same guild of pollinators, visitation patterns can lead to competition through interspecific pollen transfer (Waser, 1978, 1983), a form of reproductive interference (Armbruster et al., 1994; Muchhala & Potts, 2007). Deleterious effects of heterospecific pollen transfer include pollen lost from the donor species (less pollen available to be deposited on conspecific stigmas) (Waser, 1983; Campbell & Motten, 1985), stigma blockage (Armbruster & Herzig, 1984; Caruso & Alfaro, 2000; Brown & Mitchell, 2001) and the production of unviable seeds or unfit hybrid offspring (Morales & Traveset, 2008).
Interspecific pollen transfer by the same pollinator can be reduced in several ways. Plants may use effectively independent populations of the same pollinator species (and thus avoid competition) through separation in space or in seasonal time, or by flowering at different times of the day (Pleasants, 1980; Rathcke, 1983; Ashton et al., 1988; Stone et al., 1998). In sympatric species with overlapping flowering periods, the use of different sites of pollen placement on the pollinator's body can reduce reproductive interference amongst plant species sharing pollinators (Dressler, 1968; Armbruster et al., 1994; Muchhala & Potts, 2007; Johnson, 2010). This type of floral isolation was referred to as the ‘Pedicularis type’ by Grant (1994a). Such a mechanism can evolve only if pollen placement and stigma contact with pollinators are generally accurate and precise (Armbruster & Muchhala, 2009). Highly accurate and precise pollen placement and pollen pickup have been shown in Orchidaceae, in which pollen grains are aggregated into pollinia rather than being released separately as in most flowering plants. For example, the flowering periods of two close relatives, Platanthera bifolia (L.) L. C. Rich. and P. chlorantha (Cust.) Rchb., in northern Europe partly overlap in areas of sympatry. These two orchid species share the same moth pollinators and avoid reproductive interference by attaching pollinia to the tongues and eyes, respectively (Maad & Nilsson, 2004). In a group of 15 orchid species sharing an oil-collecting bee pollinator in Cape, South Africa, reproductive interference is reduced by the use of at least nine mutually exclusive pollinium attachment sites on the body of the bee (Pauw, 2006). Pollinium placement on the pollinator's body can be very precise and corresponds exactly to the place of stigma contact, avoiding interspecific pollination through shared pollinators. Although the Pedicularis type of floral isolation (Grant, 1994a) has been proposed to explain reproductive isolation under pollinator sharing, there have been no experimental studies to illustrate precise pollen placement and stigma contact in species with separate pollen grains, even in Pedicularis.
Pedicularis is one of the largest genera in the Northern Hemisphere, with > 200 species endemic to the east Himalayan region alone (Hong, 1983; Yang et al., 1998). The genus is characterized by great variation in the corolla, particularly in the morphology of the galea (the hoodlike upper lip of the corolla composed of two lobes fused dorsally) and the length of the corolla tube (Li, 1951; Macior, 1982; Hong, 1983; Yang et al., 1998). Sympatric Pedicularis species in the Himalayan region commonly overlap in flowering phenology in summer and are exclusively pollinated by bumblebees (Macior, 1990; Macior & Sood, 1991; Macior et al., 2001; Wang & Li, 2005; Tang & Xie, 2006; Huang & Fenster, 2007; Yang et al., 2007; Yu et al., 2012). However, hybridization in Pedicularis is considered to be rare or absent (Macior, 1982; Macior & Tang, 1997; Yang et al., 2007). In nectar-producing Pedicularis species, pollination is generally nototribic, and they have an erect upper lip (Fig. 1). Bumblebees visiting these species for nectar usually contact dehisced anthers directly and pick up pollen on the head, cervical crevice or dorsal abdomen (although bumblebees visiting them for pollen sometimes buzz to release the pollen). The nectarless species, which generally have an extended, bent or twisted upper lip, are buzz pollinated and generally sternotribic. Bumblebees visiting these species grab the curved galea and remove pollen by vibration, depositing pollen from (and picking it up on) their ventral surface (Macior, 1982, 1990; Macior & Tang, 1997; Macior et al., 2001; Tang & Xie, 2006; Huang & Fenster, 2007; Eaton et al., 2012). It has been assumed that the position of pollen deposition relates to the curvature of the galea, the tip of which positions the stigma so that it contacts the bumblebee's body at a location at which grooming is unable to remove pollen (Macior, 1982, 1990).
Although different positions of the galea tip (where the stigma protrudes) supposedly correspond to the location of pollen placement on the bumblebee's body (Macior, 1982; Grant, 1994b; Yang et al., 2007), previous observations on precise pollen placement have been conflicting and questionable. For example, the observation of sternotribic pollination in nectarless P. groenlandica Retz by Sprague (1962) was considered to be incorrect (Macior, 1968; Grant, 1994b). Macior (1968) stated that ‘the pollen falls from the basal opening of the rostrum to the lowermost petal and is scattered as a small yellow cloud enveloping the insect's body’. Yang et al. (2007) stated that ‘only those parts of a bumblebee's body that are aligned to the cleft in the corolla beak (the basal opening) can receive pollen grains’ in two nectarless Pedicularis species. These observations raise an interesting question: if pollen is released from the basal opening of the galea, but pollen is picked up from the galea tip, how can precise pollen placement be achieved in conspecific Pedicularis species?
Pollination precision would be favored to reduce interspecific pollen transfer if congeneric species share the same pollinators (Armbruster et al., 2009b). Given that the orientation of the galea dictates the position of stigma contact and pollen placement, one may predict the requirement that pollen pickup and pollen release should be in the same place on the bumblebee's body for precise pollination. Here, we test the hypothesis that different sites of pollen placement on the pollinator's body can reduce reproductive interference in sympatric Pedicularis species. We used pollen dye staining to track pollen release from the galea and fluorescent powder to detect the stigma contact site on the bumblebee's body in experimental populations involving numerous species of Pedicularis. In particular, we addressed the following questions. Where does the pollen issue from the galea – from the basal opening or the tip of the corolla beak? On which part of the bumblebee's body is the pollen placed? Which part of the bumblebee's body is contacted by the stigma?
Materials and Methods
Study species and sites
The corolla of Pedicularis consists of a corolla tube, a trilobate lower lip, an upper lip covering four introrse anthers and a style projecting from the upper lip, with the stigma protruding from the tip. The upper lip can be straight with or without teeth or shaped into a hood or helmet through the fusion of two petals (galea) with a basal opening. We studied nine sympatric species in alpine meadows at a field station, Shangri-La Alpine Botanical Garden (27°54′5″N, 99°38′17″E, 3300–3350 m a.s.l.), Yunnan Province, southwestern China from June to August in 2010. These species are representative of three major types of floral forms in Pedicularis (Li, 1951; Ree, 2005), including nectar-producing, short-tubed, non-beaked species (P. densispica Franch. ex Maxim.), nectarless, short-tubed, beaked species (P. dichotoma Bonati, P. monbeigiana Bonati, P. confertiflora Prain and P. rhinanthoides Schrenk ex Fisch. et Mey.) and nectarless, long-tubed, beaked species (P. tricolor Hand.-Mazz, P. longiflora Rudolph, P. siphonantha D. Don and P. cephalantha Franch.).
Time of pollen presentation
Pollen release relates to the pattern of pollen presentation (Harder & Thomson, 1989). To document the duration of pollen presentation, we randomly labeled 20 plants of each of the nine Pedicularis species. We picked newly opened flowers in the early morning (06:00 h) and observed the status of anther dehiscence. In our study site, bumblebees were usually active after 08:00 h on sunny days. Several flowers on each raceme or plant blooming simultaneously allowed us to record the duration of pollen presentation daily. For species in which anthers had not dehisced completely when we first checked, we examined the status of anther dehiscence every hour.
Where is pollen released from the galea?
Pollen release in nectarless species depends on the vibration or scraping of anthers (Macior, 1968, 1982; Macior & Tang, 1997), rather than direct contact with dehisced anthers as in nectared species. In the nectar-producing species P. densispica, we observed gradual anther dehiscence and pollen deposited directly on the bumblebee's head. To investigate whether the pollen falls from the gap (basal opening) or the tip of the galea in the eight nectarless species, we manipulated the galea by blocking the gap or the tip with liquid glue (Dali Group Co. Ltd, Zhejiang Province, China). An initial manipulation of flowers in a pilot experiment in the summer of 2009 indicated that different brands of liquid glue that students use to paste papers were useful in blocking the gaps and tips, and that manipulated flowers were not discriminated by bumblebees. We conducted three treatments on 10 plants of each species during July to August 2010. On each plant, three flowers were each exposed to one of three different treatments: gap blocked, with open-pollination; tip blocked, with open-pollination; or tip blocked, with bagging of flowers using small nylon nets to exclude pollinator visitation. To distinguish pollen grains from the transparent petal at the gap (the basal opening of the galea), we mixed safranin dye (1% in water) with glue to block the gap, whereas we blocked the tip with glue only. We used clean forceps to block the gap or tip with glue membrane. Three days later, we collected these flowers and counted the pollen grains that remained in the gap and in the tip of the galea under a stereomicroscope.
On which part of the pollinator's body is pollen deposited?
To examine the site of pollen placement on the bumblebee's body, we stained the pollen grains with safranin dye (1% water solution) in the three sympatric Pedicularis species (P. densispica, P. dichotoma, P. tricolor), which are representative of the three types of floral forms in the genus. The three species were abundant in the garden with thousands of individuals, and their floral phenology overlapped from late July to early August (X. Q. Shi, unpublished). In the early morning, we used clean forceps dipped into dye solution to stain pollen in dehisced anthers from c. 100 flowers of one species in a patch under a tent (2 × 2 m2). After half an hour, when the safranin dye within the anthers was dry, we removed the tent and observed pollinator visitation to the patch. After a bumblebee had visited several pollen-stained flowers (usually 2–10 flowers) in the patch, we collected it quickly by converging attack using two electric mosquito bats (Shantou Yufeng Industry Co. Ltd, Guangdong Province, China). Bees were swooned within 1 s, before they could groom pollen, by means of an electric shock. We repeated the pollen staining experiments for three, four and five sunny days in P. densispica, P. dichotoma and P. tricolor, respectively, and correspondingly collected 12, 15 and 23 visiting bumblebees. The bumblebee's body was divided into 11 parts (Fig. 2). Pollen grains from each part were picked off with gelatin cubes not containing safranin dye (Kearns & Inouye, 1993), which were then transferred to a clean slide. The slides were warmed gently to melt the jelly. We counted the pollen grains (stained red or not) from each part under a microscope.
Which part of the pollinator's body is contacted by the stigma?
To detect which part of the bumblebee's body is contacted by the stigma, we deposited greenish yellow fluorescent powder (Shandong Ji'nan Luminescent Materials Co. Ltd, Shandong Province, China) on the stigmas in the early morning in an experimental population, allowed a pollinator to visit, and then examined powder placement on the bumblebee's body. In each species, we randomly chose a patch of Pedicularis and gently coated the stigmas from 100 flowers using fluorescent powder under a 2 × 2-m2 tent. The powder was mixed with a little water, and was repeatedly added to the stigmas. Half an hour later, we removed the tent and observed pollinator visitation to the patch. The bumblebee pollinators were collected in the same way as above. This fluorescent powder experiment was repeated on three fine days in each species. The fluorescent powder glows when exposed to ultraviolet (UV) light in the dark. We took photographs of the bumblebees under UV light, and recorded the part of the bumblebee's body contacted by the stigmas.
To avoid possible mixture of the stained pollen grains or fluorescent powder among the three species in the experimental field, we manipulated one species at a time. All flowers treated with stains or fluorescent powder were quickly collected after the experiments to avoid possible environmental effects.
Single flowers usually lasted 4–7 d in the nine studied Pedicularis species. Except in the nectar-producing species P. densispica, in which four anthers dehisced gradually over 2 d, anther dehiscence in the other eight, nectarless species showed a consistent pattern of complete dehiscence. In five species (P. cephalantha, P. longiflora, P. monbeigiana, P. rhinanthoides and P. tricolor), four anthers dehisced completely in the morning; in two species (P. dichotoma and P. siphonantha), two anthers began to dehisce in the evening and pollen was available the next morning; in one species (P. confertiflora), one or two anthers dehisced completely in the morning or in the evening and the other two dehisced the next morning or evening.
Pollen release from the galea
Overall, very few pollen grains were observed at the basal gap of the galea in the three studied Pedicularis species whenever the gap or the tip was blocked, but there were thousands of pollen grains remaining at the tip when the galea gap was open and the tip was closed under open-pollination in all species (Table 1). When the tip was blocked, no pollen grains at all were observed at the gap in seven species (the exceptions were open-pollinated P. longiflora flowers, in which a few grains remained at the gap), indicating that pollen grains were released not from the gap, but from the tip. In all eight species when the tip was glued, significantly more pollen grains remained in the tip in all flowers under open-pollination than in the pollinator-excluded flowers (Mann–Whitney test, all P <0.001), indicating that pollen grains were released through the tip and pollen transfer from the anther to the tip was largely dependent on pollinator visits. After the exclusion of pollinators from these tip-blocked flowers of all species, only a few pollen grains could be observed at the tip.
Table 1. Number of pollen grains (mean ± SE) remaining at the galea gap and galea tip in eight nectarless Pedicularis species
Gap blocked + open-pollinated
Tip blocked + open-pollinated
Tip blocked + pollinator excluded
Flowers were manipulated by blocking the galea gap or the galea tip with glue, under open-pollination or pollinator exclusion with nylon nets.
41.0 ± 10.3
1237.0 ± 314.3
6400.0 ± 1117.6
0.5 ± 0.5
17.4 ± 4.8
5107.5 ± 1241.5
47.2 ± 12.3
8.7 ± 4.6
1683.0 ± 475.3
12.1 ± 4.5
30.4 ± 7.4
2522.0 ± 597.4
48.6 ± 11.2
12.7 ± 4.6
169.0 ± 47.3
5366.0 ± 1414.2
45.0 ± 10.0
7.5 ± 3.4
72.8 ± 18.6
5354.0 ± 1201.7
18.4 ± 5.1
2067.5 ± 496.5
4.6 ± 2.2
Pollen placement on the bumblebee's body
The total number of pollen grains deposited on different parts of the bumblebee's bodies was significantly different in each of the three studied species (Fig. 2; Mann–Whitney test in each species, all P <0.001). More pollen grains were attached to location 1 (head) in P. densispica, location 6M (middle of the ventral abdomen) in P. dichotoma and location 5R (right ventral thorax) in P. tricolor than to the other parts of the bumblebee's body. We observed 17 650, 1169 and 1696 safranin-stained red pollen grains from P. densispica, P. dichotoma and P. tricolor, respectively, on the bodies of nine, nine and 13 bumblebees. The red-stained pollen was also mostly placed on only one of the 11 parts of the bumblebee's body, which was different in the three species and corresponded to the same location as observed for non-stained pollen (Mann–Whitney test in each species, all P <0.001). These results showed that the sympatric species of Pedicularis each had one major position of pollen placement on the bumblebee's body that was distinct from other species. The other parts of the bumblebee's body had various numbers of pollen grains attached in the three species (Fig. 2). For example, in P. dichotoma, < 100 pollen grains were placed on the dorsal part of the bumblebee's body, fewer than three of which were red stained. However, up to 760 pollen grains were placed on one ventral body part in P. densispica and up to 250 on dorsal body parts in P. tricolor. The collected bumblebees in the three Pedicularis species were Bombus richardsi Renig (Fig. 1).
Stigma contact sites on the bumblebee's body
The experiment with fluorescent powder on the stigmas showed that powder was picked up by bumblebees and placed on different parts of the bee's body in the sympatric species. Powder was observed to concentrate on the head (location 1) in five of the seven bumblebees visiting P. densispica (Fig. 1d); in the other two bees, powder was mostly on the dorsal thorax and in the corbicula, respectively. Fluorescent powder was observed to be concentrated on the ventral abdomen (location 6M) in five of six bumblebees visiting P. dichotoma (Fig. 1e), and mostly on the ventral thorax in one bumblebee. No fluorescent powder was observed on the dorsal side of bumblebees in P. dichotoma. Fluorescent powder was observed to concentrate on the right ventral thorax (location 5R) in four of seven bumblebees visiting P. tricolor (Fig. 1f); the other three bees had powder mostly on the front legs in one bee and on both the dorsal and ventral sides in the other two bees. These observations indicate that the positions of stigmatic contact on the bumblebee's body in the three species are likely to coincide with the positions in which pollen is most likely to be placed on the bumblebee's body. Fluorescent powder was often concentrated in the corbicula of the hind legs in the three species, as a result of bumblebee grooming.
Our flower manipulations showed that, in eight nectarless Pedicularis species, pollen was released from the galea tip rather than the galea basal gap as previous observations suggested, and that pollen release was dependent on bumblebee visitation. Experimental tracking of pollen placement on the bumblebee's body (by safranin dye) and stigma contact location by fluorescent powder in the three sympatric Pedicularis species indicated that the locations of pollen placement and stigma contact were likely to coincide within each species. These locations of pollen placement and pickup differed between the three species, supporting the hypothesis of reduced interspecific pollination and the divergent use of the same pollinator (Johnson, 2010). Previous studies have proposed that precise pollen placement on the bumblebee's body would be favored to avoid reproductive interference in sympatric Pedicularis species through reduction of interspecific pollination (Macior, 1982; Grant, 1994a,b; Eaton et al., 2012), but experimental evidence is scarce. We showed that pollen release was from the galea tip by pollinator vibration. Pollen release from a small pore can limit pollen deposition directly to a small region of the bumblebee's body. This was confirmed by our measurement of the amount of pollen on 11 parts of the bumblebee's body and by safranin staining. Smaller amounts of pollen were observed on most parts of the bumblebee's body in the three species (Fig. 2). This may be attributed to vibration and grooming by the bumblebees after pollen deposition on the body, and/or variation in behavior during flower manipulation by bumblebees which may dislodge pollen grains from the original deposition location. However, the body part from which the stigma picked up the pollen was fairly specific in the three species, and this would facilitate the reduction of interspecific pollen transfer.
Floral morphology and pollinator foraging behavior influence pollination precision and accuracy (Armbruster et al., 2009a). Anther and stigma position and reward location seem to be the key factors governing pollen placement on the bodies of pollinators. For example, the diverse Neotropical genus Burmeistera exhibits wide interspecific variation in the degree of anther and stigma exsertion outside the corolla (Muchhala, 2006). Bats are the primary pollinators and the degree of exsertion determines the site of pollen deposition on the bat's head. The measurement of stigmatic pollen loads has demonstrated that the greater the difference in exsertion length between two species, the less interspecific pollen that is transferred (Muchhala & Potts, 2007). Another highly diverse genus is Stylidium in Western Australia which usually has several species flowering in sympatry. There is wide interspecific variation in nectar tube length and the position of the motile column of fused stamens and styles, which cause differential pollen deposition on the pollinator's body (Armbruster et al., 1994). The three highly diverse taxa (Burmeistera, Pedicularis and Stylidium) show a similar high degree of integration of flower structures, in particular the positions of pollen and stigma presentation, which could favour precise pollen partitioning on the pollinator's body, given that sympatric species share similar pollinators. In nectar-producing Pedicularis species, in which anthers dehisce gradually, pollen is generally dispensed onto the bumblebee's head or dorsal thorax, or via the abdomen, touching the anthers during nectar collection from the corolla tube. The main position of pollen placement depends on the position of the anthers and how deep the bee's head pushes into the corolla tube (Fig. 1a; Wang & Li, 2005; Huang & Fenster, 2007; Tang et al., 2007; Eaton et al., 2012). Our experiment of blocking the galea gap or tip indicated that pollen was removed by the bumblebee mainly from the tip rather than the basal opening (the gap) in nectarless Pedicularis species. These observations suggest that pollen falls from the small pore of the bent or twisted upper lip in nectarless Pedicularis species. This could function in the same manner as the poricidal anther dehiscence in other instances of buzz pollination in many bee-pollinated plants (Buchmann, 1983). The observation is consistent with the prediction of precise pollen placement.
Pollen placement specific to one site on the bumblebee's body has been hypothesized to have reduced reproductive interference in sympatric Pedicularis for a long time (Macior, 1982; Grant, 1994a). Our quantitative measurement of pollen amounts on different parts of the bumblebee's body has shown that pollen from each Pedicularis species has one major site, but pollen placement is not very specific in any of the three studied species, given that it covers many parts of the bee's body (Fig. 2). In many nectarless Pedicularis species, pollination is achieved through stigma contact at the site of the bee's body at which residual pollen cannot be collected by grooming (Macior, 1968, 1982; Eaton et al., 2012). If pollen grains are only placed on sites unreachable by grooming, how do bumblebees collect pollen rewards in these nectarless species? Indeed, some studies have recorded a pollen cloud seen around the bee during pollen release from the galea by bumblebee vibration (Macior, 1968; Yu et al., 2012; S. Q. Huang, unpublished), suggesting that pollen could settle imprecisely on various parts of the bee's body. Sometimes bumblebees collect only pollen in nectar-producing species (Fig. 1g). They grasp the galea edge with their mandibles, sitting in an inverted position under the galea, and rapidly vibrate (buzz) their wing muscles to shake pollen from the anthers onto their ventral surface (Macior, 1982; Harder, 1990; Robart, 2005; Tang et al., 2007). Thus, the same Pedicularis species can be pollinated nototribically and sternotribically by different behaviors of bumblebees. Furthermore, grooming behavior can transfer deposited pollen from one part of the bee's body to others. For example, in P. densispica, pollen was initially deposited on the head and dorsal thorax when bumblebees were sucking nectar, but some pollen was placed on ventral parts during the grooming of pollen from the head to the corbicula of the hind legs. In two hybridizing species Rhinanthus minor L. and R. angustifolius C.C. Gmelin (Orobanchaceae), patterns of pollen placement on the three body parts (dorsal abdomen, dorsal thorax and ventral surface) of four bumblebee species were very similar and no difference in the site of pollen placement was observed between the two sympatric species (Natalis & Wesselingh, 2012). Reproductive isolation is highly sensitive to local habitat, that is, relative frequency and spatial arrangement of species, in these two species with strong similarities in flower morphology and pollinators (Natalis & Wesselingh, 2013).
Our examination of stigma contact location on the bee's body using fluorescent powder indicated that the three sympatric species tended to have separate stigma contact sites, each corresponding to the specific pollen placement site. Each Pedicularis species is characterized by a unique corolla morphology, exemplified by the elaboration of the galea orientation in which the stigma protrudes from the tip. The diversity of galea morphology has been considered to be functionally important in positioning the contact of reproductive organs with the bodies of bumblebees (Macior, 1982; Robart, 2005; Eaton et al., 2012). Given that stigma orientation is probably more important than pollen placement on the bumblebee's body, the tremendous diversity of galea morphology in Pedicularis seems to be associated with the reduction of interspecific reproductive interference.
However, the location of stigma contact on the bumblebee's body is not 100% precise either. We noted that fluorescent powder was deposited on several parts of the bees in the three species, which may be attributed to behavioral variation of bumblebees when they are manipulating the flowers and grooming. Interspecific hand pollination among six sympatric Pedicularis species indicated that postzygotic reproductive isolation prevents the production of viable hybrids (Mao, 2010). In conclusion, our data on pollen placement and stigma contact on the bumblebee's body indicated that pollination is not sufficiently precise to completely eliminate interspecific pollen transfer, as was also found in two other species-rich genera (Muchhala, 2006; Muchhala & Potts, 2007; Armbruster & Muchhala, 2009). Bumblebees often collect pollen and/or nectar from only one Pedicularis species during one foraging bout (Macior, 1982; Adams, 1983; Yu et al., 2012). A recent study on two sympatric Rhinanthus species indicated that the foraging constancy of bumblebees changed locally with relative species abundance and spatial configurations (Natalis & Wesselingh, 2013). This possibility requires quantification in Pedicularis in further studies. Together with pollinator constancy, reciprocal pollen placement and stigma contact sites on the bumblebee's body in sympatric Pedicularis species may ultimately minimize reproductive interference.
We thank Q. Fang and M. Xie for help in the field, Z-D. Fang, Z-L. Ma and X. Hai from Shangri-La Alpine Botanical Garden for logistical support, L. Harder for valuable suggestions on experimental design, S. Corbet for drawing a model of the bumblebee's body, and W. S. Armbruster, S. Corbet, P. Wilson, R. A. Wesselingh, A. Vey and anonymous reviewers for providing valuable comments on the manuscript. This work was supported by the National Science Foundation of China (Grant nos. 31030016 and 31270281) to S-Q.H.