The E ffects of Spatial Distribution on the proportion of pollinia removed in Epidendrum radicans Orchidaceae Jennifer Woo Department of Biology Whitman College Walla Walla, Washington _______________________________________________________________________ ABSTRACT Many plants that do not offer rewards for pollination are involved in a floral mimicry. Mimic flowers copy visual cues from unrelated model species that offer rewards and rely on the chance visitation by confused pollinators. Epidendrum radicans is a mimic plant that strongly resembles two plants that provide rewards: Asclepias curassavica Asclepiadaceae and Lantana camara Verbenaceae. This study examines how spatial distri butions affect the proportion of pollinia removed PPR for the non rewarding mimic Epidendrum radicans. Nearest neighbor distances between infloresencce s, total number of flowers, and total number of pollinia removed were recorded at thirty patches of E. radicans in Monteverde, Costa Rica. A wide range of patch densities was found along with a negative relationship between proportions of pollinia removed PPR and density. Extremely large patches of E. radicans did not follow the negative trend between PPR and density. This could be due to multiple visits from naive pollinators who have not realized that no reward is offered or the possibility of multiple visits by pollinators who are reassuring themselves of the non rewarding characteristic. A patchines s index was also calculated for each patch. However, no conclusions could be made about the relationship of proportion of pollinia removed PPR and the index of patchiness. Although E. radicans is involved in a floral mimicry, the data did not suggest tha t the mimic grows in close proximity to its models. RESUMEN Muchas plantas que no of r ecen recompensa a sus poliniz adores i mitan a otras flores que si ofre cen una recompensa. Las plantas que juegan este papel copian ciertas caractersticas visuales de otras plantas y de penden del los polinizadores se confundan y las visitan casualmente. Epidendrum radicans es una planta que imita a: Asclepias curassavica Asclepiadaceae and Lantana camara Verbenaceae, dos que dan recompensa. Este estudio examin como la distribucin del espacio afecta el porcentaje de polinia que est a sacado de esta flor. La distancia del vecino m s cerca de las misma espec ie, el nmero total de las flores, y el n mero total de polinia sacado fueron a puntados en treinta p arcelas de E. radicans en Monteverde, Puntarenas, Costa Rica. Haba una gran varie dad de las densidades y se encontr una relacin negativa entre el porcentaje de polinia sacado PPR y la densidad de las plantas. Se puede explicar las visitas mltiples de los polinizadores ingenuos que no se han dado cuenta que no hay una recompensa o la posibilidad de las visitas mltiples por polinizadores que se estn asegurando a si mismos de que la flor no tiene recompensa. Las parcelas muy grandes no siguen la ten dencia negativa entre el porcentaje de pollinia quitado PPR y la densidad. Una ndice de desigualdad fue calculada por cada parcela Sin embargo, no se encontr una di ferencia significativa entre el ndice de desigu aldad patchiness" y el n mero de
poli nia sacado. Aunque E. radicans esta imitado a las otras flores, este estudio encontr que no crece muy cerca a las plantas que se esta imitando. INTRODUCTION Visual cues, olfactory cues and the presence of collectable, consumable or usable substance are all characteristics used by plants to attract pollinators Ackerman 1994. Among the vast number of plants in the world, there are some flowering plants that do not present a pollinator reward and instead use some form of deception to attract pollinators Ackerman 1994. One form of deception is floral mimicry. This idea suggests flowers without rewards mimic flowers with rewards. However, it is believed that pollinators are likely to visit the model more frequently and spend more time at a model flower than a mimic flower Haber 1984. Non rewarding plants attract pollinators by presenting the same cues that a pollinator innately recognizes Haber 1984 and its flowers are pollinated by the number of "mistake" visits due to imperfect discrim ination by the pollinators Agren 1991. Floral mimicry characteristics are frequently found among the 20,000 species in the family Orchidaceae Ackerman 1994. One example of floral mimicry in this family is Epidendrum radicans Orchidaceae. Epidendrum r adicans is a common roadside weed specie that occurs throughout the Neotropics from 1,000 to 2,000 meters Deacon 2000 and bears a simple inflorescence with flowers that vary in color from yellow to red Figure 1. After a flower has been pollinated, the lip normally a much lighter orange than the sepals and petals turns darker, becoming less attractive to pollinators Todzia 1983. It ranges from Mexico to Panama and blooms more or less through out the year Todzia 1983. Epidendrum radicans grows terre strially in dense masses on open soils, rocks, and brushy banks. Long white roots are usually found on the bases o f the newer leaves This plant is commonly known for its resemblance with two nectar producing plants: Asclepias curassavica Asclepiadaceae and Lantana camara Verbenaceae Todzia 1983. Since Epidendrum radicans offers no reward and it obtains visits by resembling other nectar producers, it is considered a Batesian mimic Bierzychudek 1981. Although E. radicans, A. curassavica, and L. camar a bear the same color flowers and attract the same butterflies, E. radi ans attracts far fewer pollinators than the model it is trying to mimic Dressier 1993. There have been only a few studies examining the distribution of all of three and how the proximity affects pollinia removal of each other. In order to compensate for the lack of a reward, E. radicans flowers possess other adaptations for infrequent visitation. Single flowers can last up to 10 days and produce as many as 0.5 million seeds. Poll en of E. radicans is packaged into single pollinia that a ttaches to the probiscis of the pollinator. This enables pollinators to visit other species without displacing the pollinia Bicrzychudek 1981 because different plants attach pollinia to different parts of the pollinator's body.
Another factor that affects the fitness of an E. r adicans plant is the spatial location between it and other conspecifics. Dense stands of E. radicans could provide the opportunity for a pollinator to realize there is no reward and leave before visiting a number of flowers Bierzychudek 1981. This idea i s supported by evidence suggesting that pollinators visiting flowers lacking rewards are less likely to visit an adjacent flower than to leave the patch Ackerman 1994. Thus, individuals in dense stands of floral mimics may not successfully reproduce. Sim ilarly, it has been suggested that larger patches of E. radicans would negatively affect pollination and the proportion of pollinia removal Deacon 2000. Nevertheless, there is reason to believe that a large patch of E. radicans may in fact increase the probability of reproductive success in certain circumstances. For instance, a larger patch may increase the number of visits by pollinators because as large patches, they are more noticeable and attractive to visitors. Consequently, an incre ase in number of visits would increase the proportion of pollinia removed. In another circumstance, the sheer increase in patch size could increase the overall number of pollinia removed, even if the proportion of pollinia removed is not affected. This stu dy will observe the spatial distribution of E. radicans flowers and examine the effect of density and patchiness on pollination success. Traditionally, researchers have focused on the female component of reproductive success by measuring fruit size and see d production Wolfe 1987. However, this study will base pollination success by quantifying pollinia removal because the pollen grains of E. radicans are packaged in a pollinaria which can be easily observed and counted Wolfe 1987. I expect the proportion of pollinia removed PPR will increase as density increases. Furthermore, patches with decreasing nearest neighbor distances or an increase in patchiness will have a positive relationship with the proportion of pollinia removed. In addition, A. cur assav ica and/or L. camara will be positively associated with Epidendrum radicans. METHODS Study site This study was conducted in Monteverde, Puntarenas Province, Costa Rica from April 9 to May 10, 2001. Natural patches of f E. radicans were located along the roadsides of Monteverde, on the road to San Luis and to Tilarn Guanacaste Province. A patch was defined as a cluster of flowering E. radicans that had a minimum distance of 10 m from the outer most inflorescence of the patch to an other E. radicans inflorescence in every direction. The nearest neighbor distance for each inflorescence was rec or ded by locating the nearest inflorescence and measuring the distance between the two in cm. The total number of flowers was quantified by counting all flowers on all inflorescences in a patch. The total number of pollinia removed was also recorded. To calc ulate the area of the patch and make comparisons of different patches possible, the
following method was used. An ellipse was drawn around the area filled with inflorescences. The radius was calculated by taking the mean of the length and width of each are a filled with inflorescences. Then a larger ellipse was drawn at a distance of 10 m from the perimeter of the area of inflorescences labeled as total area. The area of the entire ellipse was calculated as Â€r 2 where r= 10 m + the radius of the patch Figu re 2. The presence of A. curassavica and L. cama ra in flower and not in flower was also recorded for each patch. Each patch was scanned for flowers of both species by walking back and forth till the entire area of the patch had been covered. If flowers were not present, the structure of possible model plants was carefully examined. The presence of milky white latex was used as the determining characteristic of A. curassavica and a square stem was used as the determining characteristic of L. camara. In ex tremely large patches of E. radicans with more than 1,000 flowers, a sub sample was taken because of the pressure for time and efficiency. A 10 x 10 m sub patch was chosen for study based on these criteria: it seemed to be representative in terms of the av erage density of the entire patch and it was easy access. The same observations were recorded using the same methods outlined previously. In these circumstances, the total area was calculated. The total number of flowers and total number of pollinia remove d was estimated by the number of times the 10 x 10 m sub patch could fit into the area of the actual patch. An index of patchiness was determined by mean nearest neighbor distances and was calculated as m=l / 4r 2 where m= density per unit area and r= mean distance between nearest neighbor Southwood 1954. A chi squared test was used to test the association of the presence of A. curassavica and L. camara with E. radicans. RESULTS Thirty three natural patches of E. radicans were located throughout Monteverde, Costa Rica. The range of number of flowers was 1 to 258,984. The average number of flowers per pa tch was 8820 with a standard error of 8628. The calculated areas of each patch ranged 314 m 2 in a patch with a solitary inflorescence to 3896 m 2 in an extremely large patch. The average area was 627.97 m 2 with a standard deviation of 149.825. A frequency distribution of all the patches was created which showed there was a high number of a smaller patch Figure 3. In contrast, there were only a few extremely large patches with more than 201 flowers. A simple regression was used to test for correlation between the proportion of pollinia removed PPR and the density of each patch total number of flowers/area. There was no relationship found between all PPR and density values R 2 = 0.004, p value= .7261, N = 29. Nor was there a relationship between the PPR and density values whe n there was an exclusion of the PPR values of zero. In contrast, a negative relationship R 2 = 0.252, p value = 0.0400, N = 16 was found between PPR and the density when two exclusions were made: 1 the exclusion of the PPR values that equaled 0; 2 the exclusion of extremely large density Figure 4a. We can exclude the zero
PPR values by assuming that these particular patches have yet to be visited by pollinators. The patches with extremely large densities were also excluded because the numbe r of flowers and PPR could have not been accurately reflected because they were calculated by estimation. If this trend was followed, then the PPR values for extremely high densities were unexpectedly high Figure 4b. A Spearman Rank test was performed to see if there was a correlation between PPR and the index of patchiness. When the zero values for PPR were included in the test, there were no significant results N = 20, Rho corrected = 0.108, tied p value = 0.6279. The same result was calculated when the zero values were excluded from the data set N = 14, rho corrected = 0.107, tied p value = 0.6885. A chi squared was used to see whether E. radicans is positively, negatively, or not associated with A. curassavi ca and/or L. camara in flower and not in flower. Epidendrum radicans was found to not have a positive association with flowering A. curassav ica. Thus, it is non randomly dispersed without the influence of flowering A. curassavi ca X 2 = 10.8, d.f = 1. Only one plant of L. c amara was found in all of the patches of E. radicans, which did not allow for a chi squared for association to be tested. DISCUSSION The results from this study support the two different hypotheses on the effect spatial distribution on non rewarding plants. Conventional theories suggest that non rewarding plants will not be as successful in larger patches because pollinators will realize there is no reward and move on to a different plant. The second theory in this paper suggested a large patch of E. radicans could be successful because it is more noticeable and its size would expedite high pollinia removal. The negative relationship found between the PPR and density with exclusions supports the conventional theory of fl oral mimicry that denser stands of E. radicans will decrease visitation because pollinators will realize there is not a reward Bierzychudek 1981. Thus, lowering the PPR. This theory applies to the smaller patches of E. radicans whose visitors were smart enough to learn there is no reward. The data from the extremely large patches of E. radicans support the second theory that large patches can be successful because the PPR was surprisingly high and did not follow the trend between PP R and density. In this case, naÂ ve pollinators were relied upon for pollination. Most theories predict large patches will not be favored by pollinators because they do not offer a reward and it has been demonstrated that smaller, less dense patches have higher pollinia removal s uccess. However, extremely large patches of E. radicans still exist. The second theory presented in this paper explains this apparent paradox. A surprisingly high PPR is found which suggests that these large patches of E. radicans are not suffering from it s non rewarding characteristic. Figure 4b shows the two large
patches which are represented by stars. They had a higher PPR than expected, which suggests they are using different methods to attract pollinators. Large patches are more noticeable and are doi ng a better job att racting more visitors who are naÂ ve and uneducated of the non rewarding nature. This increase in number of visitors would increase in the PPR. Another benefit of a large patch is that the overall number of pollinia removed would be much greater than the number of pollinia removed in a small patch. This implies greater reproductive success compared to a small patch, even though the PPR may be lower. Further studies should examine the types of pollinators visiting the different size patches of E. radicans to determine if young or naÂ ve visitors ar e the pollin ators of the large patches and smarter older pollinators are visiting smaller patches. The frequency of visits should also be investigated. Another possibility of high PPR is that smart pollinators may require several visits to distinguish between deceitful flowers and those flowers that have been emptied by previous visitors Haber 1991. Pollinators are making multiple visits to these large patches because they want to be reassured that a reward is not offered. This also suggests that the second theory is working in the extremely large patches of E. radicans. The Spearman Rank test used see if there was a correlation between PPR and plant spacing patchiness could not lead to significant conclusions because a nearest neighbor distance could not be measured for solitary individuals. These individuals were exclu ded from the sample size, limiting it to 14. According to the conventional theory of floral mimicry, plant spacing would have a negative effect on PPR in the same manner as density. Having inflorescences closer together would limit the number of visits bec ause the smart pollinators would leave before visiting other flowers. The second theory suggests that patchiness could increase pollination success because the size w ould be more attractive to a naÂ ve pollinator. While there was no obvious relationship can be made between PPR and plant spacing for E. radicans, a study on Taxus canadensis by Allison 1990 showed that pollination success was negatively correlated with plant spacing. Allison's study does not consider the non rewarding nature of E. radicans, b ut Allison's results follow the same trend that a larger sample size would also show a negative relationship between PPR and plant spacing. Future studies investigating the relationship should include larger sample sizes. Because E. radicans relie s on the pollinators of its two models, A. curassavica and L. camara, it would be expected that E. radicans would be located near the plants or in the same patch as it models. It was determined that E. radicans is not positively associated with A. curassav i ca. This supports Deacon's theory that a non rewarding mimic should be located within the general region of its models, but not locally in the same patch. Other factors that could have affected the non positive association of E. radicans with A. curassav ica are different germination requirements or the chance that
has not permitted seed dispersal to the same areas. Further studies should look for A. curassav ica and investigate its association with E. radicans There was not enough data to make any conclus ions about the association between L. camara. Lantana cam ara is generally found at a lower altitude so it has been speculated that pollinators of L. camara of are visiting E. radicans at higher altitudes Bierzychudek 1981. This suggests that the floral mimicry is working, but the two species are not likely to be found in the same patch. The results of this study shows support the proposed theory which contradicts the conventional theory of the spatial di stribution effects on Epidendrum radicans. The new hypothesis proposes that the consequences of extremely large patches of non rewarding plants such as E. radicans may not be a negative as once proposed. These results suggest that the conventional theory s hould re evaluate its prediction of pollination success of non rewarding plants. ACKNOWLEDGEMENTS I would like to thank Karen Masters for her guidance with this study and her knowledge with Microsoft excel and Alan Masters for letting me ride the motorcycle to La Trocha. Thank you Tim Kuhman, Andrew Rodstrom, Stephanie MacFarlen, and Allison Deines for hugs, moral support, and delicious treats. Thank you Costa Rica for showing me the beauty of tropical rain forests. LITERATURE CITED Ackerman, J.D., J.A. Rodrguez and E.J. Melndez 1994. A meager offering by an epiphytic orchid is better than nothing. Biotropica 263: 44 49. Agren, J., and D.W. Schemske. 1991. Pollination by deceit in a neotropical monecious herb. Biotropica 233: 235 241. Allison, T.D. 1990. Pollen production and plant density affect pollination and seed production in Taxus canadensis. Ecology 712: 516 522. Atwood, J.T. and D.E. Mora de Re tana. 1992. Orchids of Cost a Rica. p p 1 433. The Marie Selby Botanical Gardens. Sarasota, Florida, US. Bierz ychudek, P. 1981. Asclepias, Lant ana, and Epidendrum: A floral mimicry complex? Biotropica 13 Suppleme nt: 54 58. Deacon, N. 2000. Pollinia removal and visitation in Epidendrum radicans Orchidaceae and Asclepias curassavica Asclepiadaceae CIEE Program Fall 2000 Dressl er, R.L. 1993. Field guide to the orchids of Costa Rica and Panama, pp 27. Cornell UP. New York, US.
Haber, W. 1984. Pollination by deceit in a mass flowing tropical tree Plumeria rubra L. Apocynaceae. Biotropica 164: 269 275. Todzia, C.A. 1983. Epidendrum radicans Bandera Espaola Gallito. In D.H. Janzen Ed.. Costa Rican Natural History. University of Chicago Press. Southwood, T.R.E. 1978. Ecological methods: Second edition. Chapman and Hall, New York, pp 48.