Factors influencing pollination and fruit set in Epidendrum radicans Megan Dixon Department of Biology, Northeastern University ABSTRACT reward. Flo wers who use deception may use two taxa mimicry, automimicry, or mimicry based on naivete. Epidendrum radicans has been documented to take adavantage of two taxa mimicry only in close geographic range. This study finds that although pollinia removal and pollination is low in monospecific patches of E. radicans, successful pollination is a product of pollinator naivete and/or long distance two taxa mimicry. The likelihood of pollinator visitation increases for each plant with inflorescence size. Pollinat ion rate peaks with the peak in migrating populations of naÃ¯ve butterflies. With low rates of pollination its highly likely that thriving populations rely heavily on vegetative reproduction. In the case of pollination, pollinia removal leads the plant to drop its flower early. I hypothesize that this mechanism promotes genetic diversity within the population. RESUMEN La mÃmica del engaÃ±o del alimento es un dispositivo en el que la planta atrae polinizadores con flores que no contienen una recompensa. La s flores que utilizan el engaÃ±o pueden utilizar la mÃmica de dos taxones, o automÃmica, o la mÃmica basada en la candidez. Se ha reportado que Epidendrum radicans se aprovecha de la mÃmica de dos taxones sÃ³lo en Ã¡mbitos geogrÃ¡ficos pequeÃ±os. Este estudio d eterminÃ³ que aunque la eliminaciÃ³n de los polinios y la polinizaciÃ³n sean bajas en grupos monoespecÃficos de E. radicans , la polinizaciÃ³n exitosa es un producto de la candidez polinizadora y/o la mÃmica de dos taxones a larga distancia. La probabilidad de las visitas de los polinizadores aumenta con el tamaÃ±o de la inflorescencia para cada planta. La tasa de polinizaciÃ³n alcanza el mÃ¡ximo con el pico en las poblaciones migratorias de mariposas cÃ¡ndidas. Es muy probable que las poblaciones prÃ³speras con tasa s bajas de polinizaciÃ³n dependan sobremanera de la reproducciÃ³n vegetativa. La eliminaciÃ³n del polinio lleva a la planta a botar la flor prematuramente. Se propuso que este mecanismo promueve la diversidad genÃ©tica dentro de la poblaciÃ³n. INTRODUCTION Fl owers that offer no reward get pollinated using food deception mimicry. Two taxa deceit mimicry, automimicry, and mimicry based on naivete are three described categories of this mechanism (Little, 1983). Two taxa deceit mimicry consists of two or more ta xa, a model(s) and a mimic, in which the mimic tricks its pollinator by appearing morphologically similar to its nectar producing model (Little, 1983). Automimicry is mimicry within a single species where one species (monecious or diecious) contains both m odel and mimic flowers (Little, 1983). Mimicry based on naivete is a function of pollinator mistakes by newly emerged insects or juvenile migrant hummingbirds (Little, 1983). Epidendrum radicans (Orchidaceae) offers no reward to its butterfly pollina tors and is considered an unresolved food deception mimicry by Little (1983). Automimicry is ruled
out because no flowers produce nectar or any other reward. Attempts to demonstrate multiple taxa mimicry with nectar producing Lantana camara (Verbanaceae) and Asclepias currasavica (Asclepiadaceae) have failed (Bierzychudek, 1981). Currently, it appears that multiple taxa mimicry can only be demonstrated for E. radicans inside restrictive conditions of close proximity and, at least, equal abundance (Dupre, 2004). In monospecific stands of E. radicans , multiple taxa mimicry may occur at large spatial scales if pollinators generalize amongst widely spaced monospecific patches. In this case, when pollinators enter an E. radicans patch they quickly leave, as they receive no reward from any of the flowers they visit. Similarly, when naÃ¯ve pollinators enter the patch they quickly learn that no rewards are forthcoming and abandon the patch. In either case, visits are low and restricted to the first few flow ers visited. This raises the question of how individuals of E. radicans increase the likelihood that their flowers will be the first sampled when a pollinator enters a patch. Cydista spp. (Bignoniaceae) have synchronous mass flowering episodes that a re widely spaced, assuring that all bees are naÃ¯ve (Gentry, 1974). Though E. radicans is not highly seasonal, nor exhibits synchronous mass flowering, producing more flowers at times when naÃ¯ve predators are more likely, as in the case of the arrival of m igratory pollinators, would be one way. Another might be production of larger, showier displays, as when inflorescences contain more flowers (Deacon, 2000). E. radicans flowers are known to turn color and drop with age and pollination. Here, I stud y the effects of seasonal butterfly migrants and inflorescence size on pollinia removal rates. I hypothesize that large inflorescences will attract the most pollinators and pollinia removal will increase with inflorescence size. I hypothesize that healthy plants will be able to produce larger inflorescences and thus inflorescence size with increase with plant health (total mass in leaves). If healthier plants produce larger inflorescences then removal will increase with plant health. I hypothesize that i f E. radicans uses mimicry based on naivete, the seasonal migration of butterflies will produce a peak in the fruit set as the migration signifies the arrival of naÃ¯ve pollinators. E. radicans flowers are known to turn color and drop with age and po llination (Janzen, 1983). After initial data collection, based on my observations I added a new hypothesis to my investigation that the flowers with removed pollinia will turn red and fall off earlier than flowers with intact pollinia and thus removal wil l reduce the size of the inflorescence. MATERIALS AND METHODS The study site was a patch of E. radicans along the roadside of La Trocha in Upper San Luis, Costa Rica from April 19 to May 9, 2005. The site, altitude 1350 m, is on the seasonal Pacific sid e of Costa Rica in the premontane moist forest life zone (Haber, Zuchowski & Bello, 10). Census of Inflorescence Size, Pollinia, and Seed Set For each plant (defined as a stalk with one flowering inflorescence), I counted the number of flowers on the i nflorescence, the pollinia removed, and the number of pods. I
surveyed a section of 100 plants within the patch to estimate the rate of removal and pollination in the population. Relationship Between Inflorescence Size, Plant Health, and Pollinia Removal For each plant, I counted the number of buds, flowers, flowers with removed pollinia, and leaves. When counting flowers and removed pollinia, I noted the color of the lip (red or yellow). I collected a representative leaf (of average quality and size) from each plant and measured its mass with a 10 g spring scale. I gathered data on 125 plants, 25 for each inflorescence size (1 5 flowers), to get an even representation of each size. I did not collect data on inflorescences larger than 5 flowers as the y were rare in the population and I could not obtain an appropriate sample size. Timing of Pollination and Seed Set In order to determine the rate of bud and flower production and loss, I flagged 30 plants at random and recorded the initial number of bud s and flowers on each plant. I marked the location of the youngest and oldest bud on each inflorescence with a silver paint pen ballpoint pen. Every other day for 19 days I returned to the plants to check the silver buds and recorded the number of gained (new), lost (dropped or eaten), and adjusted total in buds and flowers. I used the newest and oldest silver buds to calculate the average time between bud bir th and bloom and the average time between bloom and flower drop, respectively. I recorded the birth (date of marking) and bloom date for the youngest bud and the bloom and drop date for the oldest bud. I identified flower gain by the number of flowers wi thout blue pen marking where they attach to the inflorescence. After recording the number of new flowers, I marked all new flowers on the inflorescence with blue pen. I identified lost flowers by counting the number of scars on the inflorescence with blu e marks. After the first day, I subtracted the new total from the old total of marked scars to calculate the recent number (since the last day of data collection) of lost flowers. I identified lost buds by the number of gained flowers plus the number of bud scars. After the first day, I subtracted the new total from the old total of bud scars to calculate the recent number of dropped or eaten buds. I identified new buds by adding the new bud total to the new bud loss and subtracting the previous bud tot al from this number. Pod Position Census I randomly selected plants with. I counted and recorded the number of scars and flowers above the pod of 60 pods. I included flower number so I could accurately estimate the time since the pollinated flower bloo med. Pollinia Removal and Lip Color To determine the relationship between lip color and pollinia removal, I randomly selected 32 flowers with completely yellow lips and created 2 groups of 16, a control and an
experimental. The control group was 16 flow ers with intact pollinia. In the experimental group I removed pollinia from all 16 flowers. After two days I recorded the number of red and yellow lips in each group. To approximate the number of flowers on the inflorescence at the time of removal, I collected the number of bud scars (scars without buds on the inflorescence above the lowest bud on the stalk) and flower scars (scars above the lowest flower and below the lowest bud on the stalk) on the 125 plants from the census of inflorescence size, pollinia, and seed set. Flower scars (fs) signify early flower drop due to bud death or pollinia removal. Bud scars (bs) signify buds lost to bud death by predation or lack of resources. From the scar data I approximated the number of flowers lost to poll inia removal using the equation (fs bs). I calculated an adjusted inflorescence size (F) by adding (fs bs) to the observed number of flowers. I calculated an adjusted removal value (R) by adding (fs bs) to the observed number of removed pollinia. R ESULTS Census of Inflorescence size, Pollinia, and Seed Set From the census data I calculated an average of four pollinia removed 4/100 flowers each day. I calculated an average seed set of 1/100 flowers every 15 days. Relationship Between Infloresce nce Size, Plant Health, and Pollinia Removal Pollinia removal increased with inflorescence size (p < 0.05; Figure 1). Proportional pollinia removal increased, but not significantly, with inflorescence size (p > 0.05; Figure 2). Pollinia removal in creased with plant health but not significantly (p > 0.05; Figure 3). Inflorescence size increased with plant health but not significantly (p > 0.05; Figure 4). Number of buds increased with plant health significantly (p < 0.05; Figure 5). Of the flowers with pollinia removed, I found a significant difference between the flowers with yellow lips and the flowers with red lips (chi squared = 13, df = 1). 25% of the flowers had yellow lips and 75% had red lips. Timing of Pollination and Seed Set F rom the bud cycle chart I calculated an average bud production rate of one bud every three days and a flower production rate of one flower every four days. These averages correspond with an average loss (excluding loss to bloom) of one bud every 12 days. I calculated an average of 15 days from bud birth to bloom. The range of the bloom to death was highly variable and some plants never lost their oldest silver bud over the 19 day study. I calculated an average range/plant of 2.8 flowers in infloresc ence size over 19 days.
Pod Position Census After calculating the average bud production rate (scar production rate) and the average time from birth to bloom of buds, I was able to approximate the pollination date by estimating the bloom date of the fl owers that was pollinated. I multiplied the bud production rate by the position and added the birth to bloom time to this value. Time of pollination peaked in March, but the number of successful pollinations was not significantly different between the mo nths of pollination (chi squared = 5.25, df = 4; Figure 6). Pollinia Removal and Lip Color Using a chi squared test I found significant difference in lip color between the control and experimental group (chi squared = 13, df = 2). The control group had 12 yellow lips and four red lips while the experimental group had two yellow and 14 red lips after two days. The experiment indicates that pollinia removal significantly increases the temporal rate of change in lip color from yellow to red. It implies th at plants withdraw resources and shortly drop flowers following pollinia removal. Using a simple regression I found a significant positive correlation with adjusted inflorescence size (F) and R (p < 0.05; Figure 7). I found a significant positive corr elation with adjusted inflorescence size (F) and proportional R (p < 0.05; Figure 8). I found that F increases with R and proportional R. R 2 values were higher for Figure 7 & Figure 8 than Figure 1 & Figure 2 respectively, suggesting that the correlation with increased removal and inflorescence size is stronger than initial data collection indicates. DISCUSSION The purpose of this investigation was to determine the factors influencing pollination in E. radicans . Data from the investigation showed that p ollination is low and seed set is even lower in the patch. I found that, as predicted, larger inflorescences increase the likelihood of pollinia removal. In addition, most pods resulted from pollination a month or more before my investigation. As predic ted, fruit set peaked. Data suggested that pollinia removal was greater in the past and highest in March. Finally, I found that flowers drop quickly after pollinia removal. This investigation attempted to explain how E. radicans gets pollinated with out offering a reward. Removal and pollination in the monospecific study patch shows that they do not rely on proximate multiple taxa mimicry. Instead, E. radicans relies on pollinators initially tricked via long distance mimicry or mimicry based on naiv ete. In either case, the deception cannot be sustained, as there are no nectar producing models in the patch; only the first several flowers are likely to be visited before butterflies discover the deception. This explains the low rates of removal and po llination in the patch and indicates high intraspecific competition for the first few visits. Increasing inflorescence size is one way of out competing conspecifics in the patch. Butterflies are more likely to hit the inflorescence if it is larger, j ust by chance alone as it is a bigger target. In addition, they might actually prefer larger blooms. Data supports
that increased chance of contact increases pollinia removal with size, as there was a proportionate increase of pollinia removal in larger inflorescences. Plant health may allow for larger inflorescences, which has been shown to increase removal. A significant relationship between buds and health shows that healthy plants attempt to produce larger inflorescences. The lower correlat ion between health and inflorescence might be due to intermediate bud loss via predation or loss of resources (loss of health). The importance of naÃ¯ve pollinators to E. radicans pollinia removal is further illustrated by the timing of most pod produc tion and butterfly migration. Migrating butterfly populations, a potential source of naÃ¯ve pollinators, cross the area of my study site as they fly between the Pacific lowlands to the Atlantic side during the dry season (Haber, 1990). Haber (1990) charte d the migration of monarchs and other Costa Rican butterflies from the Pacific side to the Atlantic side during the Pacific dry season. He found a peak in the monarch migration in March, corresponding with the peak in fruit set. This correlation in fruit set and butterfly migration indicates that the arrival of naÃ¯ve migrants increases removal and pollination and that mimicry based on naivete is a successful mechanism for E. radicans . I attribute the trend in flower drop following removal to the low rate of removal and even lower rate of pollination in the population as shown by the proportion of removal and pollination census. It is more energy efficient for the plant to drop the visited flower and produce a new flower with both male and female repr oductive parts than to invest in saving a flower that only has female reproductive organs and a low probability of being pollinated. I hypothesize that this trend lends to genetic variability and out crossing. Data from the bud cycle chart indicates that inflorescence size is variable. That is, each plant has a low and high range in inflorescence size. If inflorescence size remained constant over time, pollinator behavior would favor a small proportion of the individuals with large inflorescences in the and potentially reduce genetic variability. However, ss pollinators visit the large inflorescence, the varying population is more evenly visited over time. E. radicans reproduction is low, despite great investment in flowers and reproductive parts. Deception works only occasionally yet E. radicans thrives in many disturbed areas. While pollination is a long shot that largely depends on long distance and naÃ¯ve pollinators, perhaps vegetative reproduction is responsible for a majority of the offspring . In the case of pollination, inflorescence size cycling may aid genetic variability for the population. ACKNOWLEDGEMENTS Thank you to my advisor, Alan Masters, for your time, ideas, and help. Thank you to Matt Gasner and Oliver Hyman for running erran ds, providing supplies, and answering questions. Thank you to the San Luis Eco Lodge for allowing me the use of lab space. Thank you Jessy (del Rio) Spaulding for securing the lab space. Thanks to all of my friends at the station who answered questions and made the process fun.
LITERATURE CITED Bierzychudek, 1981. Dupre, S. 2004. Evidence for floral mimicry in Epidendrum radicans with Asclepias curassavica and Lantana camara . CIEE Fall 2004, pp 73 83. Endress, P.1994. Diversity and evolutionary b iology of tropical flowers, Cambridge University Press, Cambridge. Gentry, 1974 Haber, W. 1990. Seasonal migration of monarchs and other butterflies in Costa Rica. Haber, W., W. Zuchowski, and E. Bello. 2000. An introduction to cloud forest trees: Mo nteverde, Costa Rica, 2 nd ed. Mountain Gem Publications, Monteverde de Puntarenas, Costa Rica. Janzen, D. 1983. Costa Rican natural history, The University of Chicago Press, Chicago. Little, J. 1983. A review of floral food deception mimicries with c omments on floral mutualism. Rancho Santa Ana Botanic Garden, Claremont, California.
FIGURES Figure 1. Number of pollinia removed per inflorescence increases (y = 0.208x 0.104) with the number of flowers in the inflorescenc e in a monospecific roadside patch of E. radicans . (R 2 = 0.105) Figure 2. The proportion of pollinia removed to total flower number in the inflorescence increases (y = 0.0108x + 0.134) with the number of flowers on the inflores cence in a monospecific roadside patch of E. radicans (R 2 = 0.003). Figure 3. The number of pollinia removed increases (y = 0.0132x + 0.3) with increase in total leaf mass (average leaf mass * number of leaves) in a monospecific roadside patch of E. radicans (R 2 = 0.024).
Figure 4. The total number of flowers on the inflorescence at one time increases (y = 0.0257x + 2.57) with the total leaf mass (plant health) in a monospecific patch of E. radicans . (R 2 = 0.037) Figure 5. The total number of buds on the inflorescence at on time increases (y = 0.076x + 2.67) with the total leaf mass (plant health) in a monospecific patch of E. radicans (R 2 = 0.17). Figure 6. Bars represent the total number of surveyed fruits that were pollinated for each month. Pollination for the 60 fruits surveyed peaks in March.
Figure 7. The total number of pollinia removed increases (y = 0.639x 0.94) with inflorescence size (R 2 = 0.40). Fs is flower scar and bs is bud scar. (Fs bs) estimates the number of recently dropped flowers due to recent pollinia removal. The addition of this value to observed flower and pollinia removal number adjust s the values to approximate the conditions when the pollinator visited the plant and the number of pollinia it/they removed. Figure 8. The proportion of pollinia removed to adjusted inflorescence size (F) increases (y = 0.0495 x + 0.1365) with adjusted inflorescence size, the estimated inflorescence size at the time of pollinia removal (R 2 = 0.068). (Fs bs) estimates the number of recently dropped flowers due to recent pollinia removal. The addition of this value to observed fl ower and pollinia removal number adjusts the values to approximate the conditions when the pollinator visited the plant and the number of pollinia it/they removed.
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xlink http:www.w3.org1999xlink xmlns:xsi http:www.w3.org2001XMLSchema-instance
leader 00000nas 2200000Ka 4500
controlfield tag 008 000000c19749999pautr p s 0 0eng d
datafield ind1 8 ind2 024
subfield code a M39-00358
Factores que influyen en la polinizacin y el cuajado de las frutas de Epidendrum radicans
Factors influencing pollination and fruit set in Epidendrum radicans
Food deception mimicry is a device in which plants lure pollinators with flowers that dont contain a reward. Flowers that use deception may use two taxa mimicry, automimicry, or mimicry based on naivete. Epidendrum radicans has been documented to take advantage of two taxa mimicry only in close geographic range. This study finds that although pollinia removal and pollination is low in monospecific patches of E. radicans, successful pollination is a product of pollinator naivete and/or long distance two taxa mimicry. The likelihood of pollinator visitation increases for each plant with inflorescence size. Pollination rate peaks with the peak in migrating populations of nave butterflies. With low rates of pollination, it is highly likely that thriving populations rely heavily on vegetative reproduction. In the case of pollination, pollinia removal leads the plant to drop its flower early. I hypothesize that this mechanism promotes genetic diversity within the population.
La mmica del engao del alimento es un dispositivo en el que la planta atrae a los polinizadores con flores que no contienen una recompensa. Las flores que utilizan el engao pueden utilizar la mmica de dos taxones, o auto mmica, o la mmica basada en la candidez. Se ha reportado que Epidendrum radicans se aprovecha de la mmica de dos taxones slo en mbitos geogrficos pequeos. Este estudio determin que aunque la eliminacin de las polinias y la polinizacin sean bajas en grupos monoespecficos de E. radicans, la polinizacin exitosa es un producto de la candidez polinizadora y/o la mmica de dos taxones a larga distancia. La probabilidad de las visitas de los polinizadores aumenta con el tamao de la inflorescencia para cada planta. La tasa de polinizacin alcanza el mximo con el pico en las poblaciones migratorias de mariposas cndidas. Es muy probable que las poblaciones prsperas con tasas bajas de polinizacin dependan sobremanera de la reproduccin vegetativa. La eliminacin del pollinia lleva a la planta a botar la flor prematuramente. Se propuso que este mecanismo promueve la diversidad gentica dentro de la poblacin.
Text in English.
Pollination by animals
Costa Rica--Puntarenas--Monteverde Zone--San Luis
Polinizado por animales
Costa Rica--Puntarenas--Zona de Monteverde--San Luis
Tropical Ecology Spring 2005
Ecologa Tropical Primavera 2005
t Monteverde Institute : Tropical Ecology