Pollinia Removal and Visitation in Epidendrum radicans Orchidaceae and Asclepias curassavica Asclepiadaceae Nicholas J. Deacon Department of Environmental Science, University of Iowa ABSTRACT The possibility of a floral mimicry complex between Lantana camara , Asclepias curassavica , and Epidendrum radicans sounds appealing based on their similarity in flower color and overlapping ranges in Central America. The pollinia removal in E. radicans and A. curassavica was observed under different conditions of s cale and proximity. The conditions for increased pollinia removal in E. radicans are to have a large inflorescence, close proximity to A. curassavica or L. camara in a patch, isolation from other E. radicans , occurrence in less dense patches, and avoidance of patches that contain L. camara . Asclepias curassavica benefits when alone more than when with either E. radicans or L. camara and with increasing flower number in a patch. The findings for E. radicans are consistent with what would be expected for a Ba tesian mimic with A. curassavica as a model. RESUMEN El posibilidad de una mÃmica de flores entre Lantana camara , Asclepias curassavica , y Epidendrum radicans esta atractivo porque tienen flores similar y viven en lugares similar en AmÃ©rica Central. Difere nte condiciones establecieron para observar cuando polen esta quitado de E. radicans . EncontrÃ³ que mÃ¡s polen esta quitado cuando E. radicans tiene mucho flores en una planta, esta cerca de A. curassavica o L. camara , no esta cerca de otros E. radicans , no esta en arreglos densos, y evita arreglos con L. camara . INTRODUCTION Successful pollination by animal vectors usually requires the plant to give the pollinator a reward to insure that it continues visiting other Plants of the same species. Often the rewar d is nectar and the male reproductive success of a plant is dependent upon pollen removal. Psychopilous flowers are specifically adapted to butterfly pollination. They are open during the day, have short tubes or spurs, and provide a landing platform. Many butterfly flowers contain bright contrasting colors, especially orange, red or pink Endress 1994. Plants that cheat the reward system by providing no nectar to pollinators are mimicking a rewarding species or guild. Epidendrum radicans and Asclepias cu rassavica are common weedy roadside species that occur throughout the Neotropics. Asclepias curassavica occurs from near see level to about 2,000 meters in the Neotropics and subtropics. It packages its pollen into
pollinia whose presence or absence can be seen by the naked eye. Wilson and Melampy 1983. It is typical of many flowers in having red to yellow inflorescences that offer nectar. Epidendrum radicans occurs most commonly between 1,000 and 2,000 meters. It ranges from Mexico through Central Americ a to Panama and blooms more or less throughout the year. The inflorescence is a simple, densely flowered raceme and also ranges in color from yellow to red Â€ turning more red with age Todzia 1983. It does not produce nectar and as of now, there is no kno wn reward Bierzychudek 1981. Pollinia removal in E. radicans and A. curassavica may be the result of factors operating at three different scales: within inflorescence, between plants in a patch, and between patches. A previous study examined the influen ce of inflorescence size on success of pollinia removal in A. curassavica and E. radicans . On the inflorescence level for both species, the total number of pollinia removed per inflorescence was positively correlated with inflorescence size. At the level o f the individual flower, the mean number of pollinia removed per flower did not increase with inflorescence size for A. curassavica. For E. radicans the probability of having its pollinia removed was independent of inflorescence size. It seems that for bot h species the effect of inflorescence size on pollinia removal is realized at the level of the inflorescence and not the individual flower because an individual flower has the same probability of having a pollinium removed regardless of whether it is part of a small or large inflorescence Wolfe 1987. Within patch factors that could play a role are likely to be plant and flower distribution. In a wind pollinated species, Taxus canadensis , the pollination success decreased with increasing mean nearest neigh bor distance Allison 1990. Finally between patch comparisons are dependent on the density of flowers and plant composition in different patches. Pollinia removal in experimental plots of Pleurothatllis segoviensis was higher in patches of higher density Lee 2000. Plants that present a similar display would contribute to the overall attractiveness of a patch. There has been a lot of speculation on the presence of a floral mimicry complex between Lantana camara Verbenaceae, Asclepias curassavica Ascle piadaceae, and Epidendrum radicans Orchidaceae Bierzychudek 1981. The remarkable resemblance in flower coloration and habitat type make them intriguing candidates. All three have a similar yellow, orange, and red inflorescence and are commonly found a long roadsides and on road c uts throughout the Neotropics. Lantana camara and A. curassavica are thought to be Mullerian mimics whose resemblance allows them to attract more pollinators, while E. radicans does not produce nectar so it is thought to be a Bat estian mimic of the other two Bierzychudek 1981. Because of experience with a model, pollinators visit a floral mimic expecting a reward. However, the mimic provides no reward and pollination occurs through deceit Haber 1984. While not every visit resu lts in the removal of pollinia, the number of pollinia missing from an inflorescence can be used as a relative index of visitation frequency Bierzychudek 1981. A previous study found that there were not significant differences in the number of pollinia i n stands of E. radicans alone and when it occurred with A. curassavica , nor for A. curassavica with L. camara and therefore, this was not a floral mimicry complex Bierzychudek 1981.
The purpose of this study is to examine the relationship of these three plants at different scales to determine the conditions where E. radicans and A. curassavica are losing pollinia and why. My hypothesis is that there is an effect on the pollinia removal of E. radicans and A. curassavica on three different spatial scales. METHODS The study took place over four weeks in October and November, 2000. I spent three days locating existing patches of E. radicans , A. curassavica , and L. camara . The sites were along the road from Santa Elena to Tilaran, Guanacaste Province, Costa Rica, with the farthest site being 20 kilometers North of Santa Elena. The sites consisted of steep road cuts at around 900 meters in a premontane wet forest life zone Holdridge 1967. Six sites of E. radicans alone, six sites of A. curassavica alone, one site of A. curassavica and E. radicans , one site of L. camara and E. radicans , and one site of L. camara, A. curassavica , and E. radicans were studied. The mixed patches were created by transplanting so that I could observe what was happening with E. radi cans and A. curassavica at three different levels: within plant, within patch, and between patch. A patch was separated by between 300 meters to six kilometers. More importantly though, the patches are separated by unsuitable habitat for growth of these th ree species. Within Plant To determine the effect of inflorescence size on pollinia removal, I counted the number of flowers with and without pollinia per plant. For E. radicans there is a single inflorescence and for A. curassavica there are multiple in florescences. In this case, I calculated pollinia removed for the whole plant, disregarding inflorescence number. In order to determine the effect of inflorescence size on pollinia removal, a Kendall Rank Correlation was performed. Within Patch Field Stud y Six patches of E. radicans alone and A. curassavica alone were censused for pollinia removal in each individual present at the beginning of the study. These plants were recensused after one week and then again after two weeks. To compare the pollinia lo ss between the sites a proportion of pollinia removed PPR was recorded. PPR is calculated by dividing the total pollinia available by the pollinia missing. This was performed after one week, after two weeks, and a combined calculation for weeks one to th ree. To aid in flower census calculations, pollinia from E. radicans were recorded as from either yellow or red flowers. A simple linear regression was performed on the data from E. radicans alone sites to test for a relationship between the proportion of pollinia
removed versus stem and flower density, flower number, and yellow or red flowers. The density of E. radicans was determined for all sites as well as the nearest neighbor distance in one site by measuring the total area of the site and the distance from each individual to its nearest conspecific. Butterfly Garden Study Eight days were spent at the Monteverde Butterfly Garden which has a number of greenhouses to imitate different habitats. My study was set up in the Ithomiine garden. I set up arran gements of L. camara alone, A. curassavica alone, E. radicans alone, E. radicans with A. curassavica , and E. radicans with L. camara and A. curassavica . Each day from 0930 to 1330 I recorded visits to each group. Because of the small size of the garden, th e data could be looked at as though you are looking at what goes on in a mixed patch in the field. A Friedman Test was performed to look for differences in visitation. The flower number in each group was variable so a Chi square test was used to determine if the number of visitations to a plant were different than would be expected based on the flower number in each group. Between Patch Single Species For E. radicans alone patches, the density was measured in addition to pollinia censuses to determine if d ensity was affecting the amount of pollinia being removed from different patches. This was analyzed using a simple linear regression. Effect of Other Species I collected individuals of A. curassavica and E. radicans in pots in order to create mixed patche s. One patch each of E. radicans with A. curassavica , E. radicans with L. camara , E. radicans with A. curassavica and L. camara , were created. The E. radicans and A. curassavica plants were individually marked and censused for pollinia. They were recensuse d after one week and again after two weeks. These data were analyzed using a comparison of multiple proportions test between the E. radicans alone sites and the three mixed sites. Another comparison of multiple proportions test was performed between the A. curassavica alone and mixed sites. RESULTS It should be noted that on sunny days when collecting data in the field, Anartia fatima were abundant in the patches and visiting all three species in high numbers.
Within Plant Within plant effects on pollini a removal differed somewhat between E. radicans and A. curassavica . Epidendrum radicans showed a significant increase in the number of pollinia removed with inflorescence size see Table 1; Kendall t= .205; P = .0134. However, while larger inflorescences had more pollinia removed, pollinia removal declined on a per flower basis see Table 1; Kendall t = 4.475; P <. 0001. Conversely, A. curassavica showed no increased pollinia removal with increasing number of flowers per plant see Figure 1; Kendall t= 1 .094; P = .2738 though there was a trend for plants with more flowers to have more pollinia removed. On a per flower basis, the likelihood of A. curassavica pollinia removal was independent of flower number see Figure 2; Kendall t= .287; P = .0689. Ther efore, while both species tend to benefit from larger numbers of flowers, investment in each flower yields diminishing returns in E. radicans because a lower percentage of the flowers will be visited when there are a lot of them and no additional returns i s A. curassavica . Within Patch Field Study A nearest neighbor test for distribution revealed that E. radicans has a random distribution Clark and Evans test R = 1.007. Nearest neighbor distance is not significantly correlated with pollinia removal but t he trend is that clumped individuals have more pollinia than distant individuals Figure 4; Y = .847 .042* X; RÂ²= .026; P = .1909. So, it also may be that E. radicans suffers slightly per flower in bigger patches. In addition, the random distribution of the plants concurs with this idea, since there is a trend that clumped plants suffer lower levels of pollinia removal. Butterfly Garden Study The Greenhouse experiments yielded no difference in daily visitation to the different patches once flower number was controlled see Table 2; Friedman P = .4203. There is a positive correlation between flower number in a patch and the number of visits Figure 3; Y = 23.195 + 803*X; RÂ² = .839; P = .0289. However, overall visitation showed a significant difference be tween the observed and expected visitation see Table 3; XÂ² = 145.429. The mixed patch of L. camara , A. curassavica , and E. radicans showed lower than expected visitation to L. camara , higher than expected to E. radicans , and slightly higher to A. curassa vica . Also E. radicans and A. curassavica benefit when they occur together. The visitation data from the garden says that E. radicans benefits from L. camara and A. curassavica by receiving more visits when with them than when alone. Between Patch Single Species
Epidendrum radicans showed a significant negative correlation between stem density and pollinia removal between week one and two. This suggests clumped plants had fewer pollinia removed compared to widely spaced individuals Figure 5; Y = .152 .2 1 *X; RÂ² = .684; P = .0422. However, weeks two to three and one to three indicate no significant effect of spacing on pollinia removal Y = .309 . 029*X; RÂ² = .001; P = .9450 and Y =.271 .062*X; RÂ² = .016; P = .8096. There was no significant correlation between pollinia removal and flower number in a patch for E. radicans Y week 1 2 = .14 3.499E 4*X; RÂ² = .085; P = .5744; n = 6 and Y week 2 3 = .176 + .003*X; RÂ² = .3; P = .2610; n = 6 and Y week 1 3 = .206 + .001*X; RÂ² = .204; P = .3683; n = 6. In A. curassavica the proportion of pollinia was not significantly correlated with the number of flowers in week one or week three Y week 1 = .907 + .001*X; RÂ² = .022; P=.8104 and Y week 3 = .948 .005*X; RÂ² = .153; P = .5145 but there was a significant negativ e correlation in week two Figure 6; Y = .936 .002*X; RÂ² = .872; P = .0202. This means more pollinia are removed in a patch with more flowers. Proportion of pollinia removed was also measured against flower density in E. radicans . The regressions gave no significant results Figure 7; Y week 1 2 = .151 .086*X; RÂ² = .625; P = .0613 and Figure 8; Y week 2 3 = .303 + .009*X; RÂ² = 4.096E 4; P =9696 and Figure 9; Y week 1 3 = .258 + .015*X; RÂ² = .005; P = .8940. The final measurement taken between patches was a regression between yellow and red flowers versus proportion pollinia removed in E. radicans . No significance was found between proportion pollinia removed and the number of yellow flowers for any week Figure 10; Y week 1 2 = .126 1.892 E 4*X; RÂ² =. 006; P=.8858 and Figure 11; Y week 2 3 = .142 + .008*X; RÂ² = .618; P = .0636 and Figure 12; Y week 1 3 = .25 + .001*X; RÂ² = .01; P = .8533. Insignificant results were also found when looking at red flowers Figure 13; Y = .147 .001*X; RÂ² .21; P = .3608 an d Figure 14; Y =.272 + .002*X; RÂ² =.03; P= .7416 and Figure 15; Y . 194 + .003*X; RÂ² = .421; P = .1636. Effect of Other Species When the pollinia removal in patches of E. radicans and A. curassavica alone were compared to combinations of other species, significant difference were found. Using a comparison of multiple proportions test a significant difference was found in weeks two to three in the proportion of pollinia removed in E. radicans XÂ² = 7.9288. When this week was analyzed further, it was fou nd that the E. radicans lost more pollinia with A. curassavica than with both L. camara and A. curassavica q = 7.557. Also E. radicans lost more pollinia with A. curassavica than with only L. camara q = 5.9615. There were significantly more pollinia re moved from E. radicans alone than when with L. camara and A. curassavica q = 6.0119. Lastly there were more pollinia removed from E. radicans alone than the E. radicans and L. camara patch q = 3.9024 Table 4. These data indicate that E. radicans is a s successful when alone as when with A. curassavica and it suffers when in any patch with L. camara. This comparison of multiple proportion test was also done for patches of A. curassavica alone, with E. radicans , and with both L. camara and E. radicans . Significant differences were found for week two XÂ² = 11.4462. Significantly more
pollinia were present in A. curassavica when with E. radicans than A. curassavica alone q=9.2398. More pollinia were also present when A. curassavica was with E. radicans than with L. camara and E. radicans q = 3.9892. Also more were present when with L. camara and E. radicans than when A. curassavica was alone q = 3.314. This means that A. curassavica had less pollinia removed with E. radicans than when with both L. ca mara and E. radicans and it does the best when it is alone Table 5. Week three also showed a significant difference XÂ² = 168.976. This week the pollinia present in A. curassavica were significantly more when with both L. camara and E. radicans than wit h A. curassavica alone q = 22.5320. Significantly more pollinia were present in A. curassavica when with L. camara and E. radicans than with only E. radicans q = 8.3726. Lastly, more pollinia were present in A. curassavica when with E. radicans than wh en A. curassavica was alone q = 15.504. So like week two, A. curassavica does the best when alone but its success with E. radicans was better than when with both L. camara and E. radicans , which was the reverse of the previous week Table 5. DISCUSSION With in Plant The no reward system of E. radicans allows for fewer visits since pollinators seem to figure out that there is no reward. The reward system of A. curassavica receives more visits because the pollinator is getting nectar. These data agree wi th what was found in the study by Wolfe 1987. Within Patch Pollinators leave after a few visits to a few flowers in the no nectar plants. When a reward is given, pollinators stay at the site and visit many flowers. Between Patch Single Species In E. r adicans , pollinators leave after a few visits so more flowers do not correlate to more visits. In A. curassavica , more flowers do equal more visits. Effect of Other Species When butterflies encounter no reward in E. radicans they go to other species in t he patch. They learn to recognize the difference between the reward and no reward plants which results in less visits to the cheater. The cheater helps the patch by making a better display but just increases the visits to the nectar provider.
Non reward p ollination strategies occur in other species and families in addition to Orchidaceae. Species in the Bignoniaceae that provide no reward depend on occasional visits by naÂve bees for pollination. These species exhibit a Â‚multiple bangÂƒ flowering strategy, blooming at sporadic intervals throughout the year with all flowers opening during the same few days and most of the species in the area blooming at the same time Gentry 1983. One species, Cydista diversifolia , and its sympatric congener, Cydista aequino ctialis , are very common despite rare successful pollination and fruit set. One reason is that vegetative reproduction is important in many lianas Gentry 1983. Epidendrum radicans has been proven to produce viable seeds when shelved according to Hawkes Todzia 1983. This could help to explain its success. Plumeria rubra of the Apoycnaceae provides no nectar reward to its hawkmoth pollinator. It also employs a mass flowering strategy, but for an extended blooming period. However, it lacks a specific mode l. Instead it is just a generalized mimic of other hawkmoth pollinated flowers Haber 1984. Another approach for a nectarless flower is to offer some other kind of reward. For example, it has been observed that pyrrolizidine alkaloid seeking Lepidopteran s visit Epidendrum paniculatum which, like E. radicans , gives no nectar reward DeVries and Stiles 1990. Obtaining pyrrolizidine alkaloids is important in courtship and defense of male butterflies in the subfamilies Ithomiinae and Danainae DeVries and St iles 1990. Visitation and pollinia removal are probably scale dependent. Pollinators choose flowers on a patch scale by flower and plant density and plant identity. They choose which plants to visit in a patch by the nearest neighbor distance. They choos e which flowers to visit on a plant by the reward they receive and the color of the flower. An increased density may increase the visual or olfactory stimulus to pollinators or may increase foraging efficiency Kearns and Inouye 1993 in Lee 2000. It was f ound that for Impatiens walleriana , more Hymenopterans visited larger and more dense patches Baker 1993. Therefore, non rewarding flowers should benefit from being widely spaced, since pollinators will lose interest quickly. A pollinator that visits a f lower lacking a reward is less likely to visit an adjacent flower than to move away and visit another patch Waddington 1980, 1981 in Ackerman, RodrÃguez Robles, and MelÃ©ndez. The anti predator defenses of many plants rely on a predatorsÂ„ ability to assoc iate visual stimuli with the effects of attempted ingestion Krebs and Davies 1984. It is evident that similar effects are experienced in a non rewarding plant. They want to be alone Â€ away from conspecifics and other species that offer a reward because p ollinators will learn to avoid the non rewarding species. It may require a large inflorescence per plant to assure attraction to the plant but the pollinator will not visit many of the flowers before losing interest. Alternatively, flowers that offer necta r should be clumped. Epidendrum radicans follows these non reward trends. It prefers to be alone in less dense patches and far away from other individual E. radicans . It benefits from its resemblance to A. curassavica and L. camara but only until pollinat ors learn to tell them apart and this happens most quickly when they are in the same patch. This is clear after examining the plants at different scales, since it appears that E. radicans benefits from
having the model in the region but not locally. Concl usions are mixed for A. curassavica . It prefers to be alone but this may just be because when it occurs with something else in a patch it loses visitors by competition. At times it benefits from having E. radicans in the same patch which indicates that E. radicans helps draw in pollinators but they spend more time on A. curassavica . The overall conclusions are that to increase pollinia removal, E. radicans should have a large inflorescence in order to assure at least one pollinium being removed, be near an y A. curassavica or L. camara if in the same patch, isolate from other E. radicans in a patch, possibly be in less dense patches, and avoid patches with L. camara . These conclusions seem to indicate that E. radicans does better when it is less prevalent th an the models and that it may be relying on naÃ¯ve pollinators. These requirements correlate with requirements for a Batesian mimic, the most basic being the model must be more common than the mimic Campbell 1993. In addition Janzen says that since E. rad icans contains no nectar, any visitation at all is an example of mimicry Todzia 1983. Therefore, this mimicry seems to be working. Future studies are needed in order to explore this interaction further. A single E. radicans flower lasts for about ten da ys and can produce 0.5 million seeds Bierzychudek 1981. This is a longer lifespan and more seeds than A. curassavica and L. camara produce so it could affect its fitness. Also, very little is known about what affects L. camaraÂ€s pollination success as fa r as A. curassavica and E. radicans are concerned. Gomez Pompa and M. Rausher suggest that L. camara and A. curassavica resemble each other because they are both unpalatable to herbivores Bierzychudek 1981. However, this study concludes that the resembla nce of E. radicans and A. curassavica is consistent with mimic and model interaction. A C KNOWLEDGEMENTS I greatly appreciate all the time and energy spent helping me locate my flowers and collect data by Dr. Alan Masters and Mr. Tim Kuhman. Also thank you to Mr. Jim Wolfe and the staff at the Monteverde Butterfly Garden for allowing me to perform part of my study there. LITERATURE CITED Ackerman, J.D., J.A. Rodriguez Robles and E. J. Melendez. 1994. A Meager Nectar Offering by an Epiphytic Orchid is Better than Nothing. Biotropica 261 44 49. Allson, T.D. 1990. Pollen Production and Plant Density Affect Pollination and Deed Production in Taxus canadensis. Ecology 712: 516 522. Baker, E. 1995. The Relationship of Patch Size in Impatiens walleriana to Herb ivory and Insect Visitation. UCEAP Tropical Biology Program. Fall 1993. Bierzychudek, P. 1981. Asclepias, Lantana, and Epidendrum: A Floral Mimicry Compled? Biotropica 13 Supplement: 54 58.
Campbell, N. 1993. Biology: Third Edition, pp. 1112. The Benjami n/Cummings Publishing Company, Inc. Redwood City, California. De Vries, P.J. and F.G. Stiles 1990. Attraction of Pyrrolizidine Alkaloid Seeking Lepidoptera to Epidendrum paniculatum Orchids. Biotropica 223: 290 297. Endress, P.K. 1994. Richness and Evolu tionary Biology of Tropical Flowers. Pp. 131 135. Cambridge University Press. Cambridge, Great Britain. Gentry, A.H. Cydista diversifolia Bejuco de Cuatro Filos, Jalapa. In D. H. Janzen Ed.. Costa Rican Natural History. University of Chicago Press. Hab er, W. A. 1984. Pollination by Deceit in a Mass flowering Tropical Tree Plumeria rubra L. Apocynaceae. Biotropica 164: 269 275. Holdridge, D. R. 1967. Life Zone Ecology. Tropical Science Center, San JosÃ©, Costa Rica. Lee, C. 2000. Pollination in Pleuro thallis segoviensis Orchidaceae: Color and Density Effects. Tropical Ecology and Conservation C.I.E.E. Spring 2000. Krebs, J.R. and N.B. Davies. 1984. Behavioral Ecology: An Evolutionary Approach, Second Edition, pp. 175. Sinauer Associates Inc. Sunderla nd Massachusetts. Todzia, C.A. 1983. Epidendrum radicans Bandera, EspaÃ±ola, Gallito. In D. H. Janzen Ed.. Costa Rican Natural History. University of Chicago Press. Wilson, M.F. and M.N. Melampy. 1983. Asclepias curassavica Bailarina, Mata Caballo, Mal Casada, Milkweed. In D.H. Janzen Ed.. Costa Rican Natural History. University of Chicago Press. Wolfe, Lorne M. 1987. Inflorescence size and Pollinaria Removal in Asclepias curassavica and Epidendrum radicans. Biotropica 191: 86 89.
Figure 3. Relation between the flower number in each patch to the number of visits in the Butterfly Garden. The line indicates a positive correlation between increasing number of flowers and increasing number of visits. Y = 23.195 + .803 * X; R 2 = .839; P = .839 Figure 4. Pollinia present in Epidendrum radicans increases the closer the individual plants are to each other. P = .1909
Figure 5. Proportion of pollinia removed in Epidendrum radicans from week one to two decreases with an increasing nuber of plants per patch. P = .0422 Figure 6. During week two, an increased number of flowers of Asclepias curassavica within a patch corresponded to more pollinia being removed. P = .0202
Figure 7. Trend that the proportion of pollinia removed PPR in Epiden drum radicans decreases with more flowers in a patch. P = .0613 Figure 8. Proportion of pollinia removed PPR from Epidendrum radicans shows no trend with flower density from week two to three. P = .9696
Figure 9. Proportion of pollinia removed PP R from Epidendrum radicans shows no trend with flower density from week one to three. P = .8940 Figure 10. Proportion of pollinia removed PPR from Epidendrum radicans shows no trend with the number of yellow flowers from week one to two. P = .8858
Figure 11. Proportion of pollinia removed PPR from Epidendrum radicans increases with the number of yellow flowers from week two to week three. P = .0636 Figure 12. Proportion of pollinia removed PPR from Epidendrum radicans shows no trend with th e number of yellow flowers from week one to three. P = .8533
Figure 13. Proportion of pollinia removed PPR from Epidendrum radicans shows slight negative trend with increasing number of red flowers from week one to two. P = .3608 Figure 14. Propo rtion of pollinia removed PPR from Epidendrum radicans shows no trend with the number of red flowers from week two to three. P = .7416
Figure 15. Proportion of pollinia removed PPR from Epidendrum adicans shows no trend with the number of red flow ers from week one to three. P = .1636 Table 1. Pollinia removed from Epidendru radicans . Kendall rank correlation data indicate the direction of the correlation. Number of Flowers per Inflorescence 1 2 3 4 5 6 KendallÂ„s Â… Number of plants 28 23 8 3 4 1 Number of pollinia removed 3 10 8 4 7 0 Number of pollinia removed per plant 0.11 0.43 1 1.33 1.75 .213 Percent pollinia removed 11 46 33 33 35 0 .121 Table 2. Daily visit data for Lantana camara, Epidendrum radicans, and Asclepias curassavic a in the Ithomiine section of the Monteverde Butterfly Garden. The five patches are L. Camara alone, E. radicans alone, A. curassavica alone, L. camara with E. radicans and A. curassavica , and E. radicans with A. currassavica . L. = L. camara , E. = E. ra dicans , and A. = A. curassavica . Date L. camara E. radicans A. curassavica L. E. A. E. A. 11/04/00 9 0 1 1 0 1 0 0 11/05/00 17 3 3 4 3 1 1 6 11/06/00 23 7 10 8 1 3 4 16 11/08/00 13 12 18 6 7 2 0 0 11/09/00 22 7 6 5 4 3 3 3 11/11/00 18 6 6 3 0 0 0 0 11/12/00 9 5 1 2 0 1 1 1 11/13/00 11 6 4 12 9 5 2 5
Table 3. Comparison of observed and expected number of visits to different flowers in the Butterfly Garden based on the number of flowers in each group. L. = L. camara , E. = E. radicans , and A. = A. curassavica . X 2 = 145.429 L. camara E. radicans A. curassavica L. E. A. E. A. Observed 122 46 49 41 24 16 11 31 Expected 122.4 37.4 48.62 97.58 4.76 12.24 6.12 12.24 Table 4. Between patch effects of presence of other species on E. radicans . The numbers indicate the proportion of pollinia removed, with the significant differences highlighted. L. = L. camara , E. = E. radicans , and A. = A. curassavica . The overall effect is: E = EA > EL = EAL during week two to three. Week E Alone E&A E&L E&A&L One two 0.256 0.132 0.181 0.293 Two three 0.323 > EAL q=6.119 > EL q=3.9024 0.415 >EAL q=7.557 0.211 ALE>A during week two and ALE>AE>A during week three. Week A Alone A&E A&L&E One 0.9435 0.9516 0.9516 Two 0.8874 0.9370 >ALE q=3.9892 >A q=9.2398 0.9211 >A q=11.1517 Three 0.8015 0.8973 >A q=15.5604 0.9337 >A q=22.5320 >AE q=8.3726