Pollination of Burmeistera parviflora Howell 1 The flowering biology and pollination ecology of Burmeistera parviflora (Campanulaceae: Lobelioideae) Alexandra Howell Department of Letters and Sciences, Integrative Biology University of California, Berkeley EAP Tropical Biology & Conservation, Spring 2018 8 June 2018 ABSTRACT Understanding a taxonâ€™s pollination biology and floral characteristics ca n help shed light on that groupâ€™s evolution. The genus Burmeistera (Campanulaceae: Lobelioideae) is a young but diverse genus that is almost exclusively bat pollinated. This groupâ€™s radiation has likely been caused, in part, by c hanges to floral characteristics that have evolved to reduce competition for successful pollinators. In t his study I examined the pollination biology of B. parviflora in the cloud f orests of Monteverde, Costa Rica because this species is an apparent exception to this genusâ€™ bat pollinated trend. I addressed who pollinates B. parviflora by monitoring the flowers for floral visitors, determining peak hours of nectar production, and assessing how the species â€™ morphology aligns with bat, hummingbird, or insect pollination syndromes. H ummingbirds appear to be the primary pollinator of B. parviflora. This led to the examination of whether sharing a pollinator group with bat and hummingbird pollinated B. tenuiflora has resulted in interspecific pollen transfer by addressing the rate at which interspecific and conspecific pollen transfer occurred for both species. Interspecific pollen transfer appears to occur, though it is even more rare than conspecific pollen deposition. Because of this potential for interspecific pollen transfer, I examined what morphological differences occur to reduce competition caused by interspecific pollen transfer between B. parviflora, B. tenuiflora, and a third, exclusively bat pollinated species, B. cyclostigmata. All three species had very different sets of floral traits, with traits that determine pollen or stigma placement on the pollinator differing the most between them. By illuminating B. parviflora as an other exception to Burmeistera â€™s bat pollinated trend, this study aims to improve the baseline understanding of pollination syndromes and the role they play in plant evolution, in order to better predict and understand how highelevation inhabitants, such as members of Burmeistera , may be affected in the face of changing climate patterns. La biologa de floracin y la ecologa de polinizacin de Burmeistera parviflora (Campanulaceae: Lobelioideae) RESUMEN Comprender la biologa de la polinizacin y las caractersticas florales de un taxn puede ayudar a comprender la evolucin de ese grupo. El gnero Burmeistera (Campanulaceae: Lobelioideae) es un gnero evolutivamente joven pero diverso, casi exclusivamente polinizado por murcilagos. Es probable que pa rte de la radiacin de este grupo haya sido causada por cambios evolutivos en las caractersticas florales que ayudan a reducir la competencia por polinizadores exitosos. En este estudio examino la biologa de la polinizacin de B. parviflora en el bosque nuboso de Monteverde, Costa Rica, ya que es una aparente excepcin a la tendencia de
Pollination of Burmeistera parviflora Howell 2 este gnero de ser polinizada por murcilagos. Para conocer quines son los polinizadores de B. parviflora, hice un monitoreo de los visitantes florales, determin las hor as pico de produccin de nctar y evalu cmo su morfologa se alinea con los sndromes de polinizacin de murcilagos, colibres o insectos. Los colibres parecen ser el principal polinizador de B. parviflora. Este nuevo dato condujo a explorar si el hecho de compartir un grupo de polinizadores con la especie hermana B. tenui flora (polinizada por murcilagos y colibres), causa transferencia de polen interespecfico, mediante la evaluacin de la tasa de transferencia de polen inter y conespecfico entre estas dos especies. Result que s hay transferencia de polen interespecfica, aunque es ms rara que la conespecfica. Debido a esto, examin qu diferencias morfolgicas existen para reducir la competencia causada por la transferencia de polen interespe cfico entre B. parviflora, B. tenuiflora, y una tercera especie exclusivamente polinizada por murcilagos: B. cyclostigmata. Las tres especies tienen conjuntos muy diferentes de rasgos florales, y los caracteres que ms difieren entre especies son los que determinan la ubicacin del polen o el estigma en el polinizador. Al poner en evidencia B. parviflora como otra excepcin a la tendencia de Burmeistera de ser polinizada por murcilagos, este estudio mejora la comprensin bsica de los sndromes de polini zacin y el papel que desempea en la evolucin de la planta. De esta manera, se puede comprender mejor cmo las especies de altas elevaciones, como miembros de Burmeistera , pueden verse afectados frente a los cambios en los patrones climticos . Pol lination helps increase a plant populationâ€™s genetic diversity, ensures the continuous growth of producers in an ecosystem, and helps provide food for pollinating organisms. Because plants depend on pollination to reproduce, plants often have sets of morph ological features that reflect their preferred method of pollination, whether it is abiotic, like wind or water, or biotic, like a bee, bat, or hummingbird. Collectively, these sets of plant characteristics are called a plantâ€™s pollination syndrome (Faegri & van der Pijl 1979). While some plants have generalist pollination syndromes that appeal to a wide variety of pollinators, other plants are more specialized, molding their morphology to fit the needs of one species or group of species. For example, flowers pollinated by bats will be more fragrant and less colorful, whereas flowers pollinated by hummingbirds might have no sce nt but instead have eye catching colors (Muchhala 2006, Kunz & Fenton 2005, Faegri & van der Pijl 1979, van der Pijl 1961). Tho ugh plants with more specialized pollination syndromes are reducing the number of their potential visitors, they are also preventing energy from being wasted on pollen that will not come in contact with a stigma of the same species. When a plantâ€™s morphology is better adapted to one group of pollinators and less adapted to others, this helps prevent less effective pollinators from visiting the flower and taking away pollen, nectar, or time at the flower from a more effective pollinator (Muchhala 2006). Poll inator specificity can further be increased by â€œcompetition through interspecific pollen transferâ€ (Muchhala & Potts 2007). When a group of plants shares the same group of pollinators, the plants still face problems with pollen being wasted on the wrong speciesâ€™ stigma, or from other speciesâ€™ pollen blocking their stigma. Plants reduce this competition by partitioning the pollinator resource further with adaptations that, for example, deposit pollen on different parts of a pollinatorâ€™s body or attract polli nators that are active at different times of day (Muchhala & Potts 2007, Brown & Wilson 1956, van der Pijl 1961). Similarly, even without the selective pressures of competition, differences in floral
Pollination of Burmeistera parviflora Howell 3 anatomy between individuals of the same species over tim e can result in reproductively isolated populations and facilitate speciation (Muchhala 2003). The genus Burmeistera (Lobelioideae: Campanulaceae) provides a set of examples of specialized pollination syndromes resulting from competition for a successful pollinator. It is a member of the Lobelioideae clade, a group that has undergone one of the fastest known radiations since its origin in the Andean cloud forests about five million years ago (Lagomarsino et al. 2016). Most members of the Burmeistera genus , such as B. cyclostimata , are bat pollinated and exemplify a classic chiropterophilous pollination syndrome ( Muchhala 2003, Muchhala 2006, Lagomarsino et al. 2017, van der Pijl 1961). Perhaps due in part to competition through interspecific pollen transfe r or floral traits that result in reproductive isolation, Burmeistera also has exceptions to this trend. Some species in the genus, such as B. tenuiflora, are pollinated by both bats and hummingbirds. One species , B. rubrosepala, is exclusively pollinated by the hummingbird Adelomyia malanogenys (Muchhala 2006) . B. parviflora, though its pollinator is unknown, likely represents another exception to this trend. Muchhala (2006) briefly suggests that B. parviflora is likely pollinated by bees. However, it also has many characteristics that suggest an ornithophilous pollination syndrome: it is brightly colored, but unscented and tubular (Proctor et al. 1996, van der Pijl 1961). The present study seeks to answer the following questions: (1) Who pollinates B. parv iflora ? (2) If B. parviflora is pollinated by hummingbirds, is pollen transferred between B. tenuiflora and B. parviflora? And (3) what floral differences exist between B. parviflora, B. tenuiflora , and B. cyclostigmata that could help reduce interspecific pollen transfer? METHODS Study Site I conducted this study in the Monteverde Cloud Forest Biological Reserve. I collected all data from plants and hummingbirds found along a trail that runs along the Continental Divide, using the trail as an approximately 800m long transect that ran from 1019â€™32â€ N, 8447â€™56â€ W to 1019â€™25â€ N, 8447â€™40â€ W and ranged from 1,750m to 1,800m in elevation. This region, defined by the low level clouds that are blown across the mountaintop and through the forest, has an average temperature of 18.8 C and receives, on average, 2,519mm of vertical and horiz ontal precipitation every year (Nadkarni & Wheelwright 2000). This evergreen cloud forest is also characterized by its comparatively short trees that are typically densely covered in epiphytes and its â€œdense understory of shrubs, treelets, and large herbsâ€ (Nadkarni & Wheelwright 2000, Nadkarni et al. 1995, Matelson et al. 1995). General Flowering Biology General information about Burmeistera parviflora â€™s flowering habits are not well known, but they play an important role in understanding its pollinati on biology. Like other Burmeistera , B. parviflora flowers open into the male phase and transition into the female phase (Erbar & Leins 1995). To determine how long each phase lasts, I marked thirty flower buds using flagging tape. I returned to each bud da ily and recorded when it opened, how long it took to change from male to female phase, and how long the female phase lasted before the flower senesced. In
Pollination of Burmeistera parviflora Howell 4 marking B. parviflora anthers and monitoring stigmas for interspecific pollen deposition (see below), I also collected information on the ratio of male to femalephase flowers. To determine peak hours of nectar production, in the evening, around 1700 hours, I removed and measured all of the nectar present in B. parviflora flowers. I then covered the fl owers with mesh bags to prevent nectar eating animals from accessing the flowers and affecting nectar measurements. I returned to the bagged flowers at 0600h, 0900h, 1300h, and again at 1700h to remove and measure how much nectar had been produced since th e last measurement time. I sampled nectar production by inserting a 5microliter micropipette into the flowers so the bottom of the micropipette gently touched the top of the hypanthium and the nectar was drawn up into the pipette. The height the nectar re ached in the pipette was measured and used to calculate the total nectar volume. If nectar filled the entire tube, I measured the amount of nectar, and then reinserted the tube a second time. I repeated this process until nectar ceased to appear in the tub e when inserted. I repeated this process two days in a row, monitoring about 14 flowers each da y to produce a total of about 25 nectar measurements for each time period (0600h0900h, 1000h1300h, 1400h1700h, and 1700h0600h). Pollinator of Burmeistera parviflora To determine who pollinates B. parviflora, I observed floral visits by two methods: direct observation and camera traps. I placed four camera traps at four patches of blooming B. parviflora flowers and every one to four days reviewed the camera trap data to see what they had captured. In addition to camera traps, I directly observed patches of flowers. Observational periods ranged from one to four hours between the hours of 0730 and 1230. Most observational periods were three hours long, and most occurred between 0730h and 1030h because I found nectar productio n was highest during this time. Direct observations allowed me to gather information a camera trap cannot capture, such as insect visits, and to identify hummingbird visitors to species (whi ch is only possible to do remotely with a highquality camera trap photo or video). In answering who pollinates B. parviflora, accidental observations of flower visitation (outside of formal observational periods) were also used. Interspecific Pollen Tra nsfer To determine if pollen is shared between B. tenuiflora and B. parviflora (and therefore if they share an effective pollinator), I used two different colors of fluorescent powder to cover the anthers of male B. tenuiflora and B. parviflora. I walked the length of the transect daily in search of as many B. tenuiflora and B. parviflora as possible and checked stigmas of each species for either color of powder . I then cleaned the stigma off so it would show the powder color of subsequent visits clearly. Each day, I reapplied powder to male flowers that had been marked the previous day and applied powder to newly emerged males. I counted how many st igmas were checked, what color of powder was present on a stigma (if any), and how many new males were powde red each day. Floral Characteristics I measured 20 floral characteristics of Burmeistera parviflora, B. tenuiflora, and B. cyclostigmata in order to look for possible character displacement as a result of shari ng
Pollination of Burmeistera parviflora Howell 5 pollinator guilds (Juan Morei ra, pers. comm., Muchhala 2006, Lagomarsino 2017). S ee Appendix A for a detailed description of each floral measurement. B. cyclostigmata, as an exclusively bat pollinated species, served as a reference and comparison for B. tenuiflora traits that differed from B. parviflora. Statistical analysis I used the computer language R for all analysis and graphing (R Core Team 2018). The â€œplyr,â€ â€œdplyr,â€ packages were used for some calculations, and â€œFactoMineR,â€ â€œfactoextraâ€ were used to do the Principal Com ponent Analysis (PCA) (Wickham 2011, Wickham et al. 2017, Husson et al. 2017, Kassambara & Mundt 2017). The â€œggplot2â€ package was used to visualize the data (Wickham 2009). To learn more about the lifespan of B. parviflora flowers, I took the average and standard deviation of the number of days the flowers remained in each phase. To determine the effect of time on nectar production, I did two analyses. For the first, I used the presence or absence of nectar at each f lower for each time to do a Chi Squared test to determine if time of day impacted whether a flower produced nectar. For the second, I used the volume of nectar present to do an ANOVA to examine the effect of time and flower phase (male or female) on the amount of nectar produced. I conducted a TukeyHSD test to determine which times of day had significantly higher nectar volume production. I used evidence of pollen transfer using fluorescent powders to estimate the daily rate of conspe cific and interspecific powder deposition, using the powder as a proxy for pollen. I calculated the mean for B. parviflora conspecific transfer, B. tenuiflora conspecific transfer, and interspecif i c transfer for each, resulting in four averages. As done by Muchhala (2006), to assess the overlap and differences in floral characteristics between B. parviflora, B. tenuiflora, and B. cyclostigmata, I did a principle component analysis (PCA) to reveal to what degree each speciesâ€™ floral characteristics differed from the other two species. Though measured, the size of t he subtending leaf and pedicel length played a disproportionately large role in influencing Eigenvectors relative to the role they play in the plantâ€™s pollination biology and were therefore removed for PCA. I included the top five highest Eigenvectors and their corresponding traits in this study in order to examine which traits account for the largest morphological differences between species. I also ran a MANOVA to test whether the three species differed significantly in their characteristics. I removed th e subtending leaf and pedicel length measurements for this MANOVA as well. I conducted a TukeyHSD test for each characteristic in order to see which species differed significantly in their dimensions for that trait. RESULTS General Flowering Biology Flowers remained in the male phase for 1.67 0.62 days, and in the female phase for 4.77 2.17 days. About half of all marked flowers failed to fully develop to the female phase due to herbivory, especially by a white insect larva that appeared to feed on the developing anther tube before the flower bud opened.
Pollination of Burmeistera parviflora Howell 6 Pollinator of Burmeistera parviflora The proportion of flowers that produced nectar varied significantly with tim e (Chi Squared Test, X2 = 11.287, p = 0.010, Fig. 1), and the amount of nectar produced also varied significantly with time (ANOVA, F3.3,3 = 3.33, p = 0.023, Fig. 1). Significantly more nectar was produced between 0600h and 0900h than from 1700h to 0600h (TukeyHSD, p = 0.014). The only floral visitors that I observed that appeared to have the potential to transfer pollen were hummingbirds. The camera traps did not successfully capture any visits, so all visitation events were observed in person. On occasion, I saw ants or spiders walking on the flowers, but given their sizes and locations on the flower, they did not appear to be effective carriers of pollen between flowers and I did not consider them as visitors. However, there were seemingly effective pollinator visits. For three of the four observed Burmeistera parviflora visits, the visitor was a female Purple throated Mountain Gem, Lampornis calolaemus . These three visits occurred bet ween 0830h and 1130h. The fourth B. parviflora visitor was a hummingbird that I did not see long enough to identify it to species, seen at 1715h. Additionally, on one occasion, I saw a hummingbird visit two B. tenuiflora and then, directly after, visit a B . parviflora. Therefore, interspecific pollen transfer is a possibility. Fig. 1 The graph on the left shows the percent of flowers producing nectar for each time of day. From 06000900, 67.86% of flowers were producing nectar; from 10001300, 40.00%; from 14001700, 61.54%; and overnight from 1700 to 0600, 25.93%. The graph on the left shows the average nectar production for each time of day. From 06000900, an average of 1.40 microliters of nectar was produced, from 10001300 the average was 0.77, from 14001700 it was 1.05, and overnight from 1700 to 0600 it was 0.29. The number of flowers measured for each time period for both graphs, from left (06000900) to right (17000600) are as follows: 27, 27, 25, and 13.
Pollination of Burmeistera parviflora Howell 7 Interspecific Pollen Transfer Overall, pollen deposition rates for both Burmeistera parviflora and B. tenuiflora were low (Table 1). Interspecific pollen transfer was observed once: B. parvifloracolored powder was found on a B. tenuiflora stigma. I never observed B. tenuifloracolored powder on B. parviflora stigmas. Floral Characteristics Hummingbirdpollinated Burmeistera parviflora and bat pollinated B. cyclostigmata had very different sets of morphological traits, and bat and hummingbirdpollinated B. tenuiflora also had traits that were different from both of these species (Fig. 3, Table 2). The first three Table 1 . The average daily rate of the conspecific and interspecific pollen found on B. parviflora and B. tenui flora stigmas. Species Conspecific Pollen Transfer Rate Interspecific Pollen Transfer Rate B. parviflora 0.03 8 0 B. tenuflora 0.074 0.00 9 Table 2 . The averages in millimeters for each floral trait measured for B. parviflora , B. tenui flora , and B. cyclostigmata. Overall, B. tenu i flora , and B. cyclostigmata had larger measurements than B. parviflora for all traits, except for B. tenui flora â€™s corolla spli t width and sepal width (shown in bold text ), which were both smaller than those of B. parviflora and B. cyclostigmata . Character B. parviflora B. tenu i flora B. cyclostigmata Greatest corolla length 14.39 25.42 26.43 Lesser corolla length 9.73 18.28 18.36 Greatest aperture width 2.78 13.11 17.68 Corolla split width 6.23 5.27 7.39 Sepal length 3.10 15.14 7.44 Sepal width 1.92 1.66 4.18 Pedicel length 43.25 46.05 44.32 Length of subtending leaf 37.07 85.82 43.52 Pedicel width 1.12 3.11 2.01 Tube length 6.73 13.49 11.02 Widest tube width 5.23 6.58 7.32 Narrowest tube width 2.77 3.30 3.62 Greater stigma exsertion 8.63 12.72 23.54 Lesser stigma exsertion 7.39 9.75 21.20 Anther nectar distance 14.48 26.54 36.94 Anther length 4.29 6.58 9.45 Anther width 3.15 3.75 5.90 Corolla base 5.20 6.82 7.26 Corolla midpoint width 2.84 3.41 3.72 Corolla top width 8.26 10.03 12.95
Pollination of Burmeistera parviflora Howell 8 components in the PCA analysis accounted for 65.88%, 16.90%, and 5.09% of the variation. Fig. 3. Principle Component Analysis of B. parviflora, B. tenu i flora , and B. cyclostigmata floral characteristics. Table 3 . A table displaying the p values from the TukeyHSD test done on the MANOVA for each floral characteristic between each combination of species (MANOVA, Pillai2,53 = 1.95, p < 0.001) Character B. parviflora B. tenuiflora B. parviflora B. cyclostigmata B. tenuiflora B. cyclostigmata Greatest corolla length <0.001 <0.001 0.066 Lesser corolla length <0.001 <0.001 0.989 Greatest aperture width <0.001 <0.001 <0.001 Corolla split width 0.004 0.007 <0.001 Sepal length <0.001 <0.001 <0.001 Sepal width <0.001 0.037 <0.001 Pedicel width <0.001 < 0.001 <0.001 Tube length <0.001 <0.001 <0.001 Widest tube width <0.001 <0.001 0.14 1 Narrowest tube width <0.001 <0.001 0.010 Greater stigma exsertion <0.001 <0.001 <0.001 Lesser stigma exsertion <0.001 <0.001 <0.001 Anther nectar distance <0.001 < 0.001 <0.001 Anther length <0.001 <0.001 <0.001 Anther width <0.001 <0.001 <0.001 Corolla base <0.001 <0.001 0.073 Corolla midpoint width <0.001 <0.001 0.076 Corolla top width <0.001 <0.001 <0.001
Pollination of Burmeistera parviflora Howell 9 components in the PCA analysis accounted for 65.88%, 16.90%, and 5.09% of the variation. Floral characteristic Eigenvectors with the top five highest absolute values were: anther nectar distance (0.575), greatest aperture width (0.401), greatest corolla length (0.360), greater stigma exsertion (0.333), and lesser stigma exsertion (0.285). Bu rmeistera parviflora , B. tenuiflora , and B. cyclostigmataâ€™s floral traits differ significantly (MANOVA, Pillai2,53 = 1.95, p < 0.001). Each species differed significantly from the other two species for every characteristic, barring five exceptions: B. tenuiflora and B. cyclostigmata did not differ significantly in their greatest corolla lengths, lesser corolla lengths, widest tube widths, corolla base, and corolla midpoint widths (Table 3). DISCUSSION General Flowering Biology Burmeistera parvifloraâ€™s average flower lifespan is similar to that of other previously studied Burmeistera species (Stratton 1989). Floral herbivory appeared to play a large role in the flowersâ€™ success, and more studies should be done to investigate which organisms feed on and live in B. parviflora flowers. Pollinator of Burmeistera parviflora The evidence in this study strongly suggests that Burmeistera parviflora is hummingbirdpollinated. Like many hummingbirdpollinated plants, this species produces the most nectar in the early morning (Muchhala 2003, van der Pijl 1961) . Additionally, B. parvifloraâ€™s morphology is similar to that of other hummingbird pollinated species. For example, B. parvifloraâ€™s greatest aperture width was smaller than its corolla split width, a pattern that appears in B. rubrosepala, but is reversed in B. tenuiflora , B. cyclostigmata, and other bat pollinated members of Burmeistera (Muchhala 2006). Also, like B. rubrosepala, B. parvifloraâ€™s greatest aper ture width is especially narrow â€”an average of 2.78mm, compared with the 13.11mm and 17.68mm of B. tenuiflora and B. cyclostigmata, respectively. These sets of features imply that B. parviflora may be better adapted for visitation from a pollinator with a thin beak, and less adapted to fit a batâ€™s wide face. Also, these features would mean an insect would need to be rather large to make contact with the anther, but the long narrow corolla tube would appear to only offer nectar rewards to small insects that could not make contact with the anther, so insect pollination seems unlikely (Waser & Ollerton 2006). B. parvifloraâ€™s small size reduces its â€œacoustic visibilityâ€ while its bright yellow color helps visually grab a pollinatorâ€™s attention, again suggesting hummingbird or insect pollination over bat pollination. A s insects typically rely on scent more than hummingbirds (Muchhala 2006), B. parvifloraâ€™s lack of scent suggests that its bright color is best adapted for attracting hummingbirds rather than insects . Taken together, these traits adhere well to the hummingbird pollination syndrome. This is confirmed by my visual observations of hummingbird visits to B. parviflora. Though evidence in this study cannot serve to exclude the possibility of any group of pollinators, hummingbirds appear to at least be one of the primary pollinators of this species.
Pollination of Burmeistera parviflora Howell 10 Interspecific Pollen Transfer Pollen deposition events, both conspecific and interspecific, appear to be relatively rare for B. parviflora and B. tenuiflora, even though pollination rates in this study are likely an underestimate of true pollination rates. It is possible that conspecific and interspecific pollen deposition occurred and was not detected because the pollen came from male phase flowers outside of the transect that were not marked with fluorescent powder. Despite this, from observing flowers for floral visitors, it appears that th e pollen deposition rate I detected in this study and true pollen deposition rates are not vastly different because observations of visits were infrequent. My observation of a hummingbird consecutively visiting B. tenuiflora then B. parviflora, as well as discovering fluorescent, B. parviflora powder on a B. tenuiflora stigma show that, though rare, interspecific pollen transfer is possible between these two species. One potential explanation for this low rate of conspecific and interspecific pollendepos ition is the absence of a more common or more effective pollinator for these species. As the global climate changes at an ever increasing rate, species assemblages, especially those at high elevations, are changing as well (Still et al. 1999, Lenoir et al. 2008, Colwell 2008) . For example, al though Panterpe insignis , the Fiery Throated hummingbird, has historically been abundant in Costa Rica and Panama above elevations of 1600m, this hummingbird appeared to be absent along the transect I did for this study (BirdLife International 2016, Garringues & Dean 2007). Because P. insignis and B. parvifloraâ€™s ranges should overlap, perhaps at one time, or in other locations, P. insignis was or is a common pollinator of B. parviflora. Its absence during this study may have resulted in the low observed pollination rates. Floral Characteristics As shown by the PCA (Fig. 3) and MANOVA (Table 3) , the three species of Burmeistera assessed in this study had little overlap in their floral characteristics, and those factors that differed the most are features that play an especially large role in the placement of pollen on a pollinatorâ€™s body. Of the top five Eigenvectors, three of t hem (anther nectar distance, greater stigma exsertion, and lesser stigma exsertion) directly influence the placement of pollen on the pollinator. The large differences in these particular features may be evidence of character displacement as a result of sh aring pollinator groups. At the same time, the placement of B. tenuiflora between B. parviflora and B. cyclostigmata with respect to the Dim1 axis shows that B. tenuiflora â€™s features fall somewhere between the two hummingbird and bat pollinated extremes. This is evidenced by the corolla lengths and the tube widths that did not differ significantly between B. tenuiflora and B. cyclostigmat a. The fact that the two bat pollinated species in this study share similar dimensions for these traits may be an indication that these features are related to a chiropterophilous pollination syndrome. Subject to the opposing pressures to simultaneously utilize a wider range of successful pollinators but also reduce competition through interspecific pollen transfer, it is possible B. tenuiflora developed a set of traits that achieve a balance between these two selective forces. B. tenuiflora â€™s traits are different enough from both B. parviflora and B. cyclostigmata that interspecific pollen transfer is reduced (though, as evidenced by this study, not eliminated), but not so different that it ceases to utilize bat or hummingbird pollinators.
Pollination of Burmeistera parviflora Howell 11 Conclusion As global changes in weather patterns, temperatures, and human ranges of influence alter species distributions and ecological interactions, achieving a baseline understanding of an ecosystem and its interactions help us understand and predict how climatic changes will affect interactions in the future. Understanding the flowering habits and pollination ecology of Burmeistera parviflora is especially important because, as a resident of high elevational ranges, its interactions may be subject to the most change in the coming years (Still et al. 1999, Lenoir et al. 2008, Colwell 2008). Future studies must continue to investigate the pollination biology of different plants, and other specialized ecological interactions in order to better preserve groups like this unique genus. ACKNOWLEDGEMENTS I would like to thank Frank Joyce, Federico Chinchilla, Emilia Triana, Andrs Camacho and Sof a Arce Flores for their support with all of the student projects and for making this program possible. I would especially like to thank Andres Camacho for his hel p improving my methods; tirelessly walking up the mountain very early in the morning to help me do observations and mist net hummingbirds; and for providing muchneeded moral support. I would also like to give a special thanks to Emilia Triana for translat ing the abstract to Spanish and answering all of my many questions. I would like to thank Juan Morei ra for his help finding Burmeistera , informing my methods, and providing information about what is already known about the Lobelioideae clade. Lastly, I wou ld like to thank Dennis Finger for his help with some of the statist ical analysis for this project, and the entire University of California Education Abroad Program, Costa Rica 2018 for their support, fun attitude, and help in making this program an amazin g experience. LITERATURE CITED BirdLife International. 2016. Panterpe insignis. The IUCN Red List of Threatened Species 2016. Accessed 30 May 2018. Brown,W. L. and Wilson, E. O. 1956 Character displacement. Syst. Zool. 5, 49â€“64. Colwell, R. K., G. Brehm, C. L. Cardels, A. C. Gilman, a nd J. T. Longino. 2008. Global warming, elevation range shifts, and lowland biotic attrition in the w et t ropics. Science 322:258â€“261. Erbar, C., & P. L eins. 1995. Portioned pollen release and the syndromes of secondary pollen presentation in the CampanulalesAsterales complex. Flora 190:323â€“338. Faegri K. & L. van der Pijl. 1979. The Principles of Pollination Ecology, 2nd edition. Toronto: Pergamon Pres s. Garringues R. & R Dean. 2007. The Birds of Costa Rica: A Field Guide. Cornell University Press: Ithaca, New York. Husson, F., S. Le, J. Pages, J. Josse, J. Mazet. 2017. FactoMineR: Multivariate Exploratory Data Analysis and Data Mining. R package version 1.41.
Pollination of Burmeistera parviflora Howell 12 Kassambara, A., F. Mundt. 2017. factoextra: Extract and Visualize the Results of Multivariate Data Analyses. R package version 1.0.5. < https://cran.r project.org/web/packages/facto extra/index.html> Lenoir, J., J. C. Ggout, P. A. Marquet, P. De R uffray, and H. Brisse. 2008. A significant upward shift in plant species optimum elevation during the 20th Century. Science 320:1768â€“ 1770. Kunz, T. H. & M. B. Fenton. 2005. Bat Ecology. University of Chicago Press: Chicago and London. Lagomarsino, L. P., E. J. Forrestel, N. Muchhala, and C. C. Davis. 2017. Repeated evolution of vertebrate pollination syndromes in a recently diverged Andean plant clade. Evolution 71:1970â€“1985. Lagomarsino, L. P., F. L. Condamine, A. Antonelli, A. Mulch, and C. C. Davis. 2016. The abiotic and biotic drivers of rapid diversification in Andean bellflowers (Campanulaceae). New Phytologist 210:1430â€“1442. Matelson T.J., N. Nadkarni, and R. Solano. 1995. Tree damage and annual mortality in a montane forest in Monteverde, Costa Rica. Biotropica 27:441â€“447. Muchhala, N., and M. D. Potts. 2007. Character displacement among bat pollinated flowers of the genus Burmeistera : analysis of mechanism, process and pattern. Proce edings of the Royal Society B: Biological Sciences 274:2731â€“ 2737. Muchhala, N. 2006. The pollination biology of Burmeistera (Campanulaceae): Specialization and syndromes. American Journal of Botany 93:1081â€“1089. Muchhala, N. 2003. Exploring the boundary be tween pollination syndromes: Bats and hummingbirds as pollinators of Burmeistera cyclostigmata and B. tenuiflora (Campanulaceae). Oecologia 134:373â€“380. Nadkarni, N., T.J. Matelson, and W.A. Haber. 1995. Structural characteristics and floristic composition of a neotropical cloud forest, Monteverde, Costa Rica. Journal of Tropical Ecology 11:481â€“494. Nadkarni, N. and N.T. Wheelwright. 2000. Monteverde: Ecology and conservation of a tropical cloud forest. Oxford University Press: New York. Pijl, L. van der. 1961. Ecological aspects of flower evolution. II. Zoophilous flower classes. Evolution 15:44â€“59. Proctor, M., P. Yeo, and A. Lack. 1996. The natural h istory of pollination. Timbe r Press: Portland. R Core Team. 2017. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R project.org/. Still, C. J., P. N. Foster, and S. H. Schneider. 1999. Simulating the ef fects of climate change on tropical montane cloud forests. Nature 398:608â€“610. Stratton, D. 1989. Longevity of Individual Flowers in a Costa Rican Cloud Forest: Ecological Correlates and Phylogenetic Constraints. Biotropica 21:308â€“318. Waser, N. M. and J. Ollerton. 2006. Plant Pollinator Interactions: From Specialization to Generalization. University of Chicago Press: Chicago and London. Wickham, H., R. Francois, L. Henry, K. Mller. 2017. dplyr: A Grammar of Data Manipulation. R package version 0.7.4. Wickham, H. 2009. ggplot2: Elegant Graphics for Data Analysis. Springer Verlag New York. Wickham, H. 2011. The Split Apply Combine Strategy for Data Analysis. Journal of Statistical Software, 40:129. URL ht tp://ww w.jstatsoft.org/v40/i01/
Pollination of Burmeistera parviflora Howell 13 Appendix A A description of each of the floral measurements (Juan Morei ra, pers. comm., Muchhala 2006, Lagomarsino 2017).
Pollination of Burmeistera parviflora Howell 14 Character Description C1 Greatest corolla length (base to end of lobes) C2 Lesser corolla lenght (base to D/V lobe split) W1 Greatest aperature width W2 Width at split between D/V lobes SL Sepal length SW Sepal width PL Pedicel length SubL Length of leaf subtending flower PW Pedicel width TU Tube length (base to flare in Burmeistera; base to split between ventral/lateral lobes in Centropoogn and Siphocampylus ) TW Tube width (greatest) TN Tube width (narrowest) E x1 Distal end of tube (see TU in diagram) to center of stigma (e.g. between upper and lower stigma lobes) E x2 split of dors al corolla lobes to end of anthers AN Anther nectar distance AW Anther width AL Anther length WB Width at bottom of corolla CB Corolla base WM Width at midpoint of corolla WT Width at top of corolla