Pyrrolizidine Alkaloids In Nectar: Deterrents to Generalist Pollinators and Nectar Robbers Arden G. Piland Department of Ecological and Evolutionary Biology, University of Colorado, Boulder ABSTRACT Pyrrolizidine alkaloids PAs are toxic chemicals found in multiple plant families. Eight experiments were conducted in Gardens 2 and 3 of the Monteverde Butterfly Garden to examine if PAs function as deterrents to generalist Heliconiine Nymphalidae: Heliconiinae butterflies and ant nectar thieves. T o examine if Heliconiine butterflies were deterred by PAs in nectar, cut and naturally growing flowers were manipulated to contain either a 20% sucrose solution, a solution containing 20% sucrose and PAs, a solution made with 20% sucrose and plant material that did not contain PAs, or were flushed with water to produce an empty effect. Heliconiine visits to flowers with PAs were significantly different from only sucrose. Visits to flowers containing the PA solution significantly decreased in frequency a nd duration while visits to other groups of flowers generally increased in frequency and duration. This suggests that PAs act to deter generalist butterflies. To test if PAs also deter ant nectar thieves, four different nectar solutions one with 20% suc rose solution and PAs, one made from a species closely related to the flower used to extract the PAs and 20% sucrose, one made from Lantana sp. flower heads and 20% sucrose, and one 20% sucrose solution were offered to ants on small pieces of plastic plac ed randomly around the garden. After twenty minutes the plots were surveyed and the number of ants and type of ant found on each of the four solutions were recorded for each plot. Three different species of ant visited the plots but only one species was found per plot. For all three species of ant, the solution containing PAs was the least visited. This suggests that pyrrolizidine alkaloids also act to deter potential ant nectar robbers. RESUMEN Los alcaloides de Pirrolizidina PAS son sustancias qu Ãmicas tÃ³xicas que se encuentran en mÃºltiples familias de plantas. Ocho experimentos se realizaron en los jardines 2 y 3 del JardÃn de Mariposas de Monteverde con el fin de examinar si los PAS funcionan como elementos disuasivos contra las mariposas genera listas de la subfamilia Heliconiine Nymphalidae: Heliconiinae y las hormigas ladronas de nÃ©ctar. Se manipularon flores cortadas y flores creciendo naturalmente que contenÃan ya sea una soluciÃ³n de sucrosa al 20%, una soluciÃ³n de 20% de sucrosa y PAS, una soluciÃ³n hecha de 20% de sucrosa y material vegetal que no contenÂ€a PAS, o flores que fueron lavadas con agua para producir un efecto de ÂvacÂ€oÂ‚, con el fin de determinar si las mariposas Heliconiine eran disuadidas por los PAS en el nÃ©ctar. Las visitas de las mariposas Heliconiine a las flores con PAS fueron significativamente diferentes a las visitas a las flores con sÃ³lo sucrosa. Las visitas a las flores que contenÃan la soluciÃ³n de PA disminuÃan significativamente en frecuencia y duraciÃ³n, mientras qu e en general las visitas a otros grupos de flores aumentaron en frecuencia y duraciÃ³n. Esto sugiere que los PAS actÃºan para disuadir a las mariposas generalistas. Cuatro soluciones diferentes de nÃ©ctar una con 20% de sucrosa y PAS, una hecha de una especi e estrechamente relacionada con la flor utilizada para extraer los PAS y 20% de sucrosa, una hecha con flores de Lantana sp . y 20% de sucrosa y una con 20% de soluciÃ³n de sucrosa se les ofrecieron a las hormigas en pedazos pequeÃ±os de plÃ¡stico colocados a leatoriamente alrededor del jardÃn para probar si los PAS disuadÃan a las hormigas ladronas de nÃ©ctar. Se revisaron los pedazos de plÃ¡stico despuÃ©s de veinte minutos y se registrÃ³ el nÃºmero y tipo de hormigas que se encontraron en cada una de las cuatro s oluciones. Tres especies diferentes de
hormigas visitaron los pedazos de plÃ¡stico, pero sÃ³lo se encontrÃ³ una especie de hormiga por pedazo. La soluciÃ³n que contenÃa los PAS fue la menos visitada por las tres especies de hormigas. Esto sugiere que los alca loides de pirrolizidina actÃºan tambiÃ©n como disuasorias de hormigas ladronas potenciales de nÃ©ctar . INTRODUCTION In the Tropics, the majority of plant species are rare and widely spaced. Therefore, conspecifics are often spatially isolated Bawa 1990 . Regardless, most species outcross possibly due to high levels of herbivory, competition and pathogen attack Levin 1975, Janzen 1981. Because of this, adequate long distance pollen transfer is necessary and animal pollination is favored. It is estimat ed that in tropical lowland rainforests, 98 99% of the flowering plants rely on animals for pollination Bawa 1990. As a result, tropical plants have produced various ways to attract pollinators to visit their flowers, including energy rich nectar Simps on and Neff 1983. However, many nectars consist of more than simple sugars as energy rich rewards. The nectar of many plant species contains complex and sometimes toxic secondary compounds Baker and Baker 1975, Kearns and Inouye 1993. In tropical low land forests alone Baker 1978 found that phenolics were present in 50% of nectars and that 12% contained alkaloids. Although much is still unknown about the function of these components, it has been proposed that they reduce nectar robbery. Janzen 19 77 suggested that ants rarely visit flowers because either floral tissue or nectar contains chemicals distasteful or harmful to ants. Ghazoul 2001 found that 18 of 25 ant species were repelled by floral acacia chemicals and provided support for Janzen s theory. Toxic substances in nectar might also inhibit legitimate pollinators. However, iridoid glycosides found in the nectar of Catalpa speciosa Bignoniaceae were found to deter potential nectar thieves while having no effect on bumblebees, carpente r bees or sphingid moths which act as legitimate pollinators of the species Stephenson 1982. Generalist pollinators can be a problem for spatially isolated plants. Such plants increase their fitness when they limit visits to fewer species Rhoades and Bergdahl 1981. Because pollen and nectar are expensive, generalist pollinators that are less likely to visit a conspecific flower are costly and lower a plant s fitness. It is possible that toxic nectar constituents act to promote fewer pollinator spec ies and, hence, greater pollinator constancy Rhoades and Bergdahl 1981, Baker and Baker 1982. Pyrrolizidine alkaloids PA s are toxic secondary compounds found in the nectar of plants belonging to the families Fabacea, Apocynaceae, Asteraceae, Boragina ceae, and Compositae Brown et al. 1991. Ithomiine Nymphalidae:Ithomiinae and danaine Nymphalidae: Danainae butterflies have been shown to be attracted to flowers with nectar that contains these chemicals Masters 1968, Pilske 1975, Boppre 1978. Th ey sequester PAs and utilize them as pheromones for sexual attraction and as a defense against predators Brown 1984, Masters 1990. PAs seem to act as an attractant to ithomiines and danaines while inhibiting more generalist butterfly foragers. A study by Masters 1991 provided preliminary support for the theory that PA s not only attract specialist pollinators but also deter generalist pollinators. He found that Agraulis vanillae Nymphalidae: Heliconiinae a generalist pollinator, consumed less of a sucrose solution containing a PA, monocrotaline, than a solution of pure sucrose. He also reported that the butterflies learned to avoid artificial flowers containing PA solution.
However, Landolt and Lenczewski 1993 failed to support PA induced inhib ition in Heliothis virecens tobacco budworm moth, Trichoplusia ni cabbage looper moth and Agraulis vanillae gulf fritillary butterfly. For this specific toxic nectar constituent, there is still debate regarding its function. This study attempts to further test the role of PAs in plant nectars. I propose that in addition to attracting ithomiines, pyrrolyzidine alkaloids will function to limit visitation of generalist pollinators, like those belonging to the Heliconiinae subfamily, while deterring ne ctar robbing ants. METHODS Study Site and Duration Observations were made daily starting from 8am to 12pm for three weeks April 19 th to May 8 th in Gardens 2 and 3 of the Monteverde Butterfly Garden, Monteverde, Costa Rica. The Butterfly Garden is loc ated in the premontane wet life zone at 1340 meters. However, individual gardens have been constructed to imitate different elevations and environments throughout Costa Rica. Garden 2 has been constructed to mimic a mid elevation canopy environment 500 10 00m and houses species of butterflies belonging to the sub family Heliconiinae Nymphalidae. Plants common to Monteverde grow naturally in the garden and include Lantana camara Verbenaceae, Stachytarpheta frantzii Verbenaceae, and Impatiens walleria na Balsaminaceae. Four jars of freshly cut L. camara inflorescences are suspended at two meters above the ground along a center path in the garden, as well. These cut inflorescences act as feeders and are filled daily with a 20% sucrose solution. Two to four different species and fifteen to twenty five individuals were present at any one time during the observations. Garden 3 is a reproduction of a mid elevation 500 1000m understory environment and contains butterflies belonging to the sub family I thomiinae Nymphalidae. Two to four species and five to ten individuals were present at the time of study. I. walleriana is planted in this enclosure and two separate platforms hold jars of cut inflorescences of L. camara and Ageratum rugosa Asteraceae that are changed regularly. The L. camara acts as a feeder and is filled daily with 20% sucrose solution. Study Subjects It is currently believed that twenty four of seventy recognized species of Heliconiine butterflies inhabit Costa Rica and range from sea level to 2,800m on both slopes DeVries 1987. Commonly known as Longwings , most species have preferred larval host plants belonging to the Passifloraceae family. They are long lived and survive for several months Goode 1999. Their diet cons ists of both pollen, usually from vines in the Cucurbitaceae family, and nectar from a large number of flower species. In contrast to their loyal pollen collecting behavior, most have been shown to visit a variety of flowers to obtain nectar, and act as g eneralist pollinators Baker and Baker 1975; DeVries 1987. The two most abundant species found in Garden 2 were Heliconius erato and Dryas iulia both of which visit a wide variety of flowers foraging for nectar DeVries 1987. Ithomiine butterflies occu r from Mexico through Central and South America and in Costa Rica are commonly found in all habitat types from 600 1500m DeVries, 1987. The larval hostplants for these butterflies are mainly in the Solanaceae and Apocynaceae
family. They primarily feed on nectar from plants that produce PAs, but females occasionally feed on bird droppings as a nitrogen source Ray and Andrews, 1980. Most species migrate along an elevational gradient and males commonly form leks and use pheromones derived from pyrrozyl idine alkaloids to attract females Haber 1978. PA Extraction and Formation of Nectar Solutions Two closely related species from the Asteraceae family Ageratum rugosa and Ageratum pinchensis are common in Monteverde. A. rugosa produces pyrrolizidine al kaloids and A. pinchensis does not Bartzat 2000. Extracts from both species were used to ensure that findings were not a result of nectar viscosity or consistency. To prepare these solutions, 5.5 g of fresh flower heads were picked along the road leadi ng up to the Biological Station. The flowers were ground with a mortal and pestle in 30mL of methanol. The plant material was filtered and the solutions were placed in a hot water bath to evaporate the methanol, leaving PAs if present and other plant c onstituents. 30 mL of a 20% sucrose solution were added to the extracts. 20% sugar solutions were made periodically throughout the experimental phase. Solutions were kept refrigerated when not in use. Treatments were administered by injecting the solut ions with syringes into the corollas of flowers. For experiments seven and eight, fresh solutions were made. Nine grams of flower heads were picked and calyces removed to avoid the possibility of chemicals in vegetative parts contaminating solutions. An extract using Lantana sp. was also created to be used as an additional control nectar. The same methods for extraction, formation and administration were used. Pyrrolizidine Alkaloids as Pollinator Determinants Experiments 1 7 In order to test the at tractive and deterrent nature of pyrrolyzidine alkaloids to different pollinators, both subfamilies of butterflies were exposed to cultivated Lantana spp. flowers that were manipulated to contain different nectar solutions. EXPERIMENT 1a: Establishing Co lor Preferences for Ithomiine and Heliconiine butterflies Because floral traits such as color could be a confounding factor that influences attractiveness Morris 2005, Endress 1994, Appanah 1990, Hannan 1981, a preliminary study was done to determine but terfly color preferences. Three patches of different colored potted, cultivated Lantana were placed equidistantly from each other on the ground in both gardens 2 and 3. The plots were of equal size aprox. 25cm x 25 cm. Each plot was composed solely of one color yellow, purple, white and all flowers were filled with a 20% sucrose solution. Observations of which flowers visited were recorded in one hour increments. Observation periods alternated between gardens for four hours each of the two days. EX PERIMENT 1b: Color Choice of Heliconiine butterflies for Cultivated Impatiens To examine if the Heliconiine butterflies would exhibit a clear color choice for a different species of flower, experiment 1a was repeated with three color morphs of cultivated Impatiens . The color choices included red, pink and white flowers. All were treated with a 20% sucrose solution, observations were made and visits recorded for two, one hour increments.
EXPERIMENT 2: Location Effects and Nectar Preferences for Heliconii ne and Ithomiine butterflies Four hanging jars of fresh cut Lantana camara in garden two were filled with sucrose solution. The plots were observed for two, one hour periods to establish if Heliconiine butterflies had a preferred feeding location. Three jars of L. camara were placed in garden 3 and the same observations were made for Ithomiine butterflies. On day two in garden 2, the preferred location was filled with the PA nectar solution, two other locations were treated with sucrose solution and the third with the nonPA producing plant nectar solution referred to as nonPA. In garden 3, one plot contained PA nectar, one sucrose and one nonPA nectar. Both garden plots were observed and visits to each feeder were recorded for two, one hour increment s for two days. EXPERIMENT 3/4: Location and Color Choice on Heliconiine Visitation Rates for Cultivated Lantana . Potted, cultivated Lantana flowers from experiment 1 were placed in three wire baskets and hung from the ceiling next to the hanging jar s in Garden 2 to see if a suspended location would increase visitation. Native, cut Lantana camara normally in the feeders were removed from the jars to limit foraging choices. Potted Lantana of three colors were filled with the sucrose solution. Observ ations were made for one hour. The baskets were then taken down and flowers were cut from each of the three different color morphs and placed in three of the four feeders, to see if this had any effect on visitation frequency or color choice. All were re filled with the sucrose solution. These plots were again observed for one hour. EXPERIMENT 5: Lack of Nectar and PAs on Flower Visitation Rates for Heliconiine Butterflies. Yellow cultivated Lantana were placed in two of the hanging feeders in Garden 2. Another two were filled with native orange, L. camara . All hanging flowers were filled with sucrose solution. Natural L. camara, I. walleriana, and S. frantzii grow in the garden. Visits to each of these different feeders and flower types were recorde d for one hour. After an hour, the hanging L. camara were flushed with water to simulate empty flowers. Visits were recorded again for one hour. On day two of this experiment, the PA solution was added to the hanging L. camara and visits were recorded for an hour. During hour two, a PA solution was added to the L. camara growing naturally in the garden and visitation rates were observed for an hour. EXPERIMENT 6: Lack of Nectar and PAs on Visit Duration and Frequency of Flower Visitation for Helic oniine Butterflies. Flowers for this experiment were the same type in experiment 5 hanging cultivated Lantana and L .camara, L. camara natural , I. walleriana, and S. frantzii . During phase one of the experiment, both species of hanging Lantana cul tivated yellow and orange L. camara and bushes of L. camara were filled with sucrose solution. The number of visits and the length of visit to each flower type were recorded for one hour. Length of visit measured with a stopwatch was defined by three categories: short 0 3 sec., medium 4 7 and long greater than 8 seconds. Observations were made for one
hour. Sucrose solution was then reapplied and a second round of observations were recorded. This phase was repeated for two days. For phase tw o, corollas of both hanging L. camara and natural L. camara were flushed with water to make them empty for the first hour. Hanging yellow cultivated Lantana was filled with a sucrose solution. During hour two, sucrose solution was added to the previous ly empty L. camara . Number of visits and duration of visit were recorded for all five flower types in the garden. Observations were taken for one day only. Phase three consisted of placing PA nectar in the natural and hanging L. camara and sucrose solu tion was injected into the yellow cultivated hanging variety for the first hour. The corollas of the flowers containing PA solution were then flushed with water and sucrose solution was reapplied to all three Lantana groups. Observations were recorded f or an additional hour. This treatment was performed for two days. Differences in rates of flower visitation and duration between hours for each phase were calculated to determine if the presence of PAs or the lack of nectar had an effect . EXPERIMENT 7: Role of Pyrrolizidine Alkaloids as Ant Deterrents To determine if pyrrolizidine alkaloids function as a deterrent to nectar robbers, different nectar solutions were offered to ants in Garden 2 of the Monteverde Butterfly Garden. Twenty 5x5cm plastic squa res were placed on the ground randomly throughout the garden. Each plot consisted of 4 choices of solutions: 20% sucrose and PA flower extract, 20% sucrose and non PA flower extract, 20% sucrose and L. camara solution, and a 20% sucrose solution see abov e for extraction details. 1mL puddles of each solution were placed randomly on each plastic plot. The plots were left for twenty minutes and then surveyed for number and type of ant present at each solution for each of the twenty plots. This was repeat ed for four days. RESULTS EXPERIMENT 1a: Establishing Color Preferences for Ithomiine and Heliconiine Butterflies Four hours of observation in each garden, resulted in just nineteen visits for the Heliconiine species and ten for the Ithomiine butterflie s. For the Heliconiine butterflies, purple flowers were the most frequently visited with 9 visits, yellow had intermediate visitation 7 visits and white were the least frequented 3 visits; Figure 1a. Low number of visits to these flowers was in stark contrast to the high frequency of visits to freshly cut, hanging L. camara that the Butterfly Garden keeps in constant supply for the butterflies. For this reason, subsequent studies utilized the hanging flowers for testing visitation. Ithomiine butter flies preferred white flowers 4 visits, and purple and yellow flowers were equally frequented 3 visits each; Figure 1b However, there was no significant difference in visitation rates between the three color morphs for neither the Heliconiine X 2 =2.947 , df=2 nor Ithomiine X 2 = 0.199, df=2 butterflies and thus no clear color preference was evident.
a 0 2 4 6 8 10 purple white yellow Flower Color Number of Visits b 0 1 2 3 4 5 purple white yellow Flower Color Number of Visits FIGURE 1.a. Number of Heliconiine butterfly visits to different colors of cultivated Lantana in Garden 2 of the Monteverde Butterfly Garden over a span of four hours. Heliconiine butterflies visited purple flowers most frequently, yellow somewhat frequently and frequented white flowers the least. However, the butterflies did not appear to have a clear preference for one color flower over another X 2 =2.947, df=2. b. Number of Ithomiine butterfly visits in Garden 3. They visited white most frequently and yellow and purple less. These butterflies did not exhibit a clear color preference either X 2 = 0.199, df=2. EXPERIMENT 1b: Color Choice of Heliconiine Butterflies for Cultivated Impatiens During two hours of observations the Heliconiine butterflies were observed visiting the flowers seven times. They appeared to prefer red 4 visits over white 3 visits or pink 0 visits; Figure 2. However, rates of visitation for the different color morphs were not significantly different X 2 =3.714, df=2. Low visit frequencies suggest that the butterflies were uninterested, wary or otherwise unlikely to visit flowers on the ground. This was in sharp contrast to the frequent visits to the hanging feeders during the same period.
0 1 2 3 4 5 Pink white red Color Number of Visits FIGURE 2. Observations of the number of visits by Heliconiine butterflies to three different color morphs of cultivated Impatiens flowers in Garden 2 of the Monte verde Butterfly Garden over a two hour period. Red appeared to be the preferred color but no significant differences were found between visitation rates % 2 =3.714, df=2. EXPERIMENT 2: Location Effects and Nectar Preferences for Heliconiine and Ithomiin e Butterflies 84 visits were recorded over a two hour time period on Day 1 for Heliconiine butterflies and 13 visits were recorded for Ithomiine butterflies. Heliconiine species showed locational preferences among the four foraging sites X 2 =14.571, df=3 . Location two accounted for 38% 32 out of the 84 of the visits. Location one was the second most frequented 26 visits, 31%, followed by locations four 17 visits, 20% and three nine visits, 11%; Figure 3a, blue bars. Ithomiine butterflies visit ed locations one and two of three feeders equally six visits and location three less frequently one visit over the two hour period Figure 3b, blue bars. Ithomiine butterflies did not show a significant preference for one location over another X 2 =3. 846, df=2. During day two Heliconiine butterflies were observed 59 times and Ithomiine butterflies were recorded visiting the flowers 5 times over a two hour time span. With the addition of PA solution to location two and nonPA solution to location four the relative proportion of visits to each location changed. Heliconiine butterflies visited each location at a more even frequency than for day one and the locational bias of location two that was seen the day before dramatically decreased X 2 =8.764, df= 3; Figure 3a, purple bars; X 2 values compare the relative frequencies of Day 1 with the relative frequencies of Day 2, expected values were derived from the proportional frequencies observed during the sucrose treatment of Day 1.With the addition of PAs t o the preferred feeder, visits to this location decreased and visits to location three with sucrose only previously the least preferred site increased. Thus, PAs changed feeder preferences such that the preferred location decreased in total number of vi sits.
The Ithomiine butterflies were observed only four times visiting flowers on day two. After the PA solution was added to location two and nonPA solution was added to location three, the butterflies visited location two the most 2 times and locatio ns one and three equally once over the two hour period Figure 3b, purple bars. However, these rates of visitation did not show a great change from those observed during day number one X 2 =0.499, df=2. Due to the consistently low visitation rates, It homiine butterflies were not used in subsequent experiments. a. Heliconiine 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 1 2 3 4 Location Proportion of Total Visits Day 1 Day 2 b. Ithomiine 0 0.1 0.2 0.3 0.4 0.5 0.6 1 2 3 Location Proportion of Total Visits Day 1 Day 2 FIGURE 3. Observations of foraging location preference in Gardens 2 and 3 of the Monteverde Butterfly Garden. Values are shown as proportion of total visits for each day. Blue Bars represe nt Day 1 of the experiment in which all locations contained flowers filled with sucrose solution. The purple bars represent Day 2 of the experiment where solutions containing PAs were added to the preferred site as determined from the day before, and no nPA solution was added to another location. 3a shows that on Day 1, Heliconiine butterflies preferred location two X 2 =14.571, df=3, but during Day 2 the location effect greatly decreases and all locations are frequented more evenly. Visitation to locati on two which contained PAs decreased and location three sucrose only increased after the treatments were administered X 2 =8.764, df=3. 3b shows that during day one Ithomiine butterflies did not have a clear locational preference X 2 =3.846, df=2. Al so, during day two there was not a great change in visitation rates after the treatments were added X 2 =0.499, df=2. EXPERIMENT 3/4: Location and Color Choice on Heliconiine Visitation Rates for Cultivated Lantana . Four visits to feeders with diffe rent colors of cultivated Lantana were observed for the first hour. As figure 4a shows, most visits three were to the yellow variety, followed by the white one and no visits were recorded for the purple. The Heliconiine butterflies did not seem to pr efer any particular color more than the other X 2 =3.501, df=2. Similar findings were observed when the flowers were cut and placed in the hanging jars X 2 =4.66, df=2. Yellow flowers were again visited most often six times,
but during this hour visits to purple flowers two were greater than visits to white flowers one; Figure 4b. This was again in stark contrast to the natural Lantana camara bushes, which received dozens of visits during the same period. a 0 0.5 1 1.5 2 2.5 3 3.5 Yellow purple white Color Number of Visits b 0 1 2 3 4 5 6 7 yellow purple white Color Number of Visits FIGURE 4. Observations of Heliconiine visits to a hanging, potted Lantana flowers of different colors and b cut Lantana flowers in hanging jars, each over a one hour time span. Yellow was the most frequented color in both situations, but significant differences in visitation rates were no t found between colors a.X 2 =3.501, df=2; b.X 2 =4.66, df=2 EXPERIMENT 5: Absence of Nectar and PAs on Flower Visitation Rates for Heliconiine Butterflies. Day 1: During the first hour of observations, Heliconiine butterflies were recorded visiting all of the flowers in the garden 65 times. The most frequented group was the natural L. camara bushes growing in the garden 33 times, 51%, followed by the cut, hanging L. camara 17 visits, 26%. Low visitation rates were recorded for the yellow hanging La ntana four visits, 6%, I. walleriana five visits, 8% , and S. frantzii 6 visits, 9% ;Figure 5, blue bars. When the hanging L. camara were emptied during hour two, there was not a significant change in visitation rates to any of the flower groups X 2 =3.27, df=4. Out of 49 recorded visits, the natural L. camara were still the most frequented 22 visits, 45%, then the hanging L. camara 11 visits, 22%, and visitation was still low to the yellow hanging Lantana five visits, 10%, I. walleriana four visits, 8% , and S. frantzii seven visits, 14%; Figure 5, purple bars. Day 2: In the two hours of observation, 116 visits were recorded. With the addition of a PA solution to the hanging L. camara, the distribution of visitation rates significa ntly differed from what would be expected based on the relative distribution of visits to feeders from the sucrose treatments on Day 1 X 2 =14.164, df=4. The natural L. camara were again the preferred foraging choice, but to an even higher degree; 65 of t he visits 56% were to these flowers. The hanging L. camara still received visitors 16 visits, 14%, but much fewer than would be expected based upon rates from day 1. Visitation rates significantly increased for the hanging yellow Lantana 14 visits, 12% and to a lesser degree for the S. fantzii 12 visits, 10%; Figure 6.
For hour two with the addition of the PA solution to both the hanging and the natural growing L. camara, a significant change in visitation rates from what would be expected from Day 1, was again observed X 2 =21.253, df=4. The visitation rates to the hanging and natural growing L. camara decreased while visits to the yellow Lantana which contains a sugar solution and to the S. frantzii increased Figure 7. 0 0.1 0.2 0.3 0.4 0.5 0.6 L. camara yellow Lantana natural S. frantzii I. walleriana Flower Type Proportion of Total Visits Hour 1 Hour 2 FIGURE 5. Prop ortion of total visits of Heliconiine butterflies to different types of flowers over a two hour time period in Garden 2 of the Monteverde Butterfly Garden. Blue bars represent the first hour where sucrose solutions were added to L. camara and the yellow L antana . The butterflies showed a clear affinity for the naturally growing and hanging L.camara . Purple bars represent hour two when the hanging L. camara was flushed with water to make them empty . No significant change in visitation rates resulted fro m this treatment X 2 =3.27, df=4
0 0.1 0.2 0.3 0.4 0.5 0.6 L. camara yellow Lantana natural S. frantzii I. walleriana Flower Type % of Total Visits Day 1 Day 2 FIGURE 6. Visitation rates for Heliconiine butterflies to different flower types containing different nectar solutions. Values are shown and percent of total visits. The blue bars correspond to data from Day 1 of the experiment where both the hanging L. camara and the yellow Lantana contained sucrose solutions and are to provide a reference point to compare the effects of adding a PA solution to the hanging L. camara Day 2, purple bars. There was a significant change in visitation rates to the different flower types after the addition of the PA solution X 2 =14.164, df=4. Visits to the hanging L. camara decreased and visits to the yellow Lantana increased. 0 0.1 0.2 0.3 0.4 0.5 0.6 L. camara yellow Lantana natural S. frantzii I. walleriana Flower Type % of Total Visits Day 1 Day 2 FIGURE 7. Proportions of visits by Heliconiine bu tterflies to different flower types during a one hour interval. Blue bars represent visitation rates from Day 1 of the experiment where both the hanging L. camara and the yellow Lantana contained sucrose solutions and are to provide a reference point to c ompare the effects of adding a PA solution to the hanging and natural growing L. camara Day 2, purple bars. Visitation
rates after the addition of the PA solution changed significantly X 2 =21.253, df=4. Visits to both of the flower groups containing P As decreased as visits to the yellow Lantana and S. frantziii increased. EXPERIMENT 6: Lack of Nectar and PAs on Visit Duration and Frequency of Flower Visitation for Heliconiine Butterflies. Due to few observations of short visits 0 3 seconds, short and medium 4 7 seconds length visit categories were combined for the purpose of statistical analyses. The duration times will be referred to as long greater than 8 seconds and short 0 7 seconds. Also, because expected values were calculated using o bserved rates of visitation from the hour of sucrose treatment in each phase, any observed values of zero during this observational period were defined as one visit to allow for chi squared analyses to be performed . For the first phase, 162 visits were r ecorded for the first hour of observation hour 1=data of the first hour for two days combined. The majority of visits were of long duration to the natural 50 long visits, 31% of total visits and hanging 48 long visits, 30% L. camara . 17 visits we re recorded for the yellow Lantana 15 long, 9%; 2 short, 1%. The butterflies tended to stay at both the S. frantzii and the I. walleriana for shorter time periods 10 short visits, 6%; 19 short visits, 12% respectively vs. one long visit for both; Figur e 8, purple and blue bars. Hour two of this phase, when the flowers were refilled with sucrose solution, yielded very similar results. When visitation rates for the two hours were compared, no significant differences in abundance of duration of visits t o the different flower types were found between hours X 2 =5.422, df=4. For this second hour 109 visits were recorded. Again, the most visits were long, and to both the natural 31 long, 28% vs. 6 short 6% and hanging 36 long, 33%; 2 short, 2% L. cama ra . The majority of visits to the yellow Lantana were of long duration 10 long, 9% vs. 2 short, 2% whereas the visits to both the S. frantzii and the I. walleriana were generally shorter 8 short, 7% vs. 2 long 2%; 12 short, 11% vs. 0 long; Figure 8, y ellow and teal bars. For phase two, observations from hour two when the flowers were injected with sucrose solution were very similar to the findings from phase one. Long visits to the natural 22 long, 31% vs. 1 short, 1% and hanging 19 long, 27% vs. 1 short, 1% L. camara were most frequent. The majority of visits to the yellow Lantana were also long 7 long, 10% vs. 1 short, 1%. For the S. frantzii 8 visits 11% were long and 2 short 3%. For I. walleriana six visits 8% were short and 4 6% were long Figure 9, yellow and teal bars. This shows that Heliconiine butterflies have a strong preference for the native, orange L. camara . Because this trend is consistent between phases it is not due to daily preferences or time of day. However, when PAs were present in both the hanging and natural L. camara the duration of visits to different flower types changed significantly X 2 =324.415, df=4. Out of the 101 visits recorded, the most frequented flowers were still the natural L. camara 28 vi sits however, the duration of the visits differed from those observed for the sucrose treatment period. Out of the 28 visits, 17 were short 17% of total visits and 11 11% were long. The same trend occurred in the hanging L. camara . Out of the 16 vi sits, 13 13% were short and 3 3% were long. Visit length increased for S. frantzii 21 long, 21% and 0 short and I. walleriana 14 long, 14% and 10 short, 10%. Long visits to S.
frantzii accounted for 20% of all visits for both duration lengths w hich was the highest frequency of visits for any on flower type and time length. Long visits to the yellow hanging Lantana slightly increased as well from 9% of total visits during hour one to 11% after the PAs were added; Figure 9, purple and blue bars . 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 L. camara yellow Lantana natural S. frantzii I. walleriana Flower Type % of Total Visits mediumhr1 Longhr1 mediumhr2 Longhr2 FIGURE 8. Proportion of visits and length of visits by Heliconiine butterflies to different flower types for two different observational periods in Garden 2. To ensure that percent of visits are not an artifact of time of day, sucrose treatments we re implemented for both hours. This phase of the experiment is to act as a control for subsequent phases. For the natural and hanging L. camara , and the yellow Lantana , visits were generally of long duration. The opposite was found for S. frantzii and I . walleriana . Most of visits to these flower type were of short duration. No significant difference was found between hours one and two in terms of duration or distribution of visits % 2 =5.422, df=4 suggesting that the relative frequencies and durations of visits are not caused by time of day.
0 0.05 0.1 0.15 0.2 0.25 L. camara yellow Lantana natural S. frantzii I. walleriana Flower Type % of Total Visits medium PA period LongPA period mediumsucrose Longsucrose FIGURE 9. Differences in frequency and duration of visits of Heliconiine butterflies to different flowers types during two different flow er manipulation periods. Values are shown as percent of total visits for the hour observation period. Yellow and blue bars represent the period of time when sucrose solutions were present in both types of L. camara and the yellow hanging Lantana . Visits were mostly of long duration biased towards both types of L. camara . Purple and blue bars depict visitation length and frequency for the phase when PAs were present in both types of L. camara . Duration and frequency of visits during this phase were sign ificantly different from the trends observed during the sucrose phase X 2 =324.415, df=4. Visits to flowers containing PAs switched from being long greater than 8 seconds to short 0 8 seconds. Visits to other flowers during this time were generally o f long duration. For phase three, the results from hour one only sucrose treatments were consistent with the trends observed during the first two phases. Natural and hanging L. camara were again frequented the most and the visits were primarily long in duration 9 long 26% vs. 5 short 14% and 6 long 17% vs. 3 short 9%, respectively. Visits to the yellow Lantana were mostly long 3 long 9%, 1 short 3% as well as to the S. frantzii 3 long 9% vs. 1 short 3%. Visits to the I. waleriana were equally long and short 2 of each 6%; Figure 10. When the flowers were emptied visit lengths significantly differed from those that would be expected based on the first hour of observations X 2 =12.284, df=4. However, the natural and hanging L . camara both were still the most frequented and mostly by long visits 17 long 31% vs. 8 short 15% and 5 long 9% vs. 4 short 7% respectively. The largest change was for the S. frantzii which showed an increase in short visits 5 short 9% vs. 4 long 7%. The yellow Lantana again had more long visits 5 long 9% vs. 0 short, and visits to the I. walleriana were again equally long 3 visits, 5% and short 3 visits, 5%; Figure 10.
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 L. camara yellow Lantana natural S. frantzii I. walleriana Flower Type % of Total Visits mediumempty Longempty mediumsucrose Longsucrose FIGURE 10. Proportion of visits to different types of flow ers by Heliconiine butterflies at the Monteverde Butterfly Garden. Yellow and Teal bars represent hour one when all of the Lantana spp . were injected with sucrose solution. Visits are mostly dominated by long visits, and the natural and hanging L. camara were the most commonly visited. Blue and purple bars correspond to the phase where the natural and hanging L. camara were empty. Visits were significantly different X 2 =12.284, df=4, however trends were similar. The biggest source of change was visits to the S. frantzii became shorter. EXPERIMENT 7: Role of Pyrrolizidine Alkaloids as Ant Deterrents 74 plots total plots consisting of three different species of ant were observed. Only one species of ant was found on a given plot. Big Black BB ants were present on 61 of the plots, Medium reddish brown MRB were on six plots and tiny red TR ants were observed on 3 plots. For every ant type, the average number of individuals found visiting the four nectar offerings was always lowest for the offer ing that contained PAs Figure 11. There were significant differences between number of individuals found at each nectar offering Friedman test BB, X 2 =100.991, p<0.0001, MRB, X 2 =28.08, p<0.0001, TR, X 2 =9.30, p=0.0256.
a -2 0 2 4 6 8 10 12 14 Sugar NonPA Lantana PA Nectar Type Average number of visits b. 0 5 10 15 20 25 30 Sugar non-PA lantana PA Nectar type Average number of visits
c 0 20 40 60 80 100 Sugar non-PA lantana PA Nectar type Average number of visits FIGURE 11. Graphs of the average number of ants visiting each nectar solution. Graph a shows results for the big black ants, number of visits to each were significantly different Friedman X 2 =100.991, p<0.0001, df=3. B shows significant differences for the medium reddish brown ant type Friedman X 2 =28.08, p<0.0001, df=3. C shows averages for the tiny red any type, average number of individuals at each solution were significantly different Friedman X 2 =9.30, p=0.0256, df=3. The general trend for each ant type shows that ther e were always fewer individuals visiting the solution that had PAs. Ants did not seem to prefer the non PA solution either. The sugar solution and the solution made from Lantana sp. were always the most visited. DISCUSSION The data suggest that pyrroli zidine alkaloids act to both deter generalist pollinators in the sub family Heliconiinae Nymphalidae and to deter at least three species of ant nectar robbers. As Figures 3a and 6 suggest, when PAs are present, visits to the manipulated flowers decrease and visits to other flowers increase. Other treatments such as the flowers being empty generally did not produce significant trends. The presence of the non PA solution in experiment two, where identical feeders of orange L. camara were treated with dif ferent nectar solutions, did not seem to have an effect on visitation rates. This assures that the observed trends are not artifacts of the butterflies responding to simple differences in the flowers and validates the suggestion that the PAs are responsib le for the observed changes in visitation rates. Visits to the flowers containing PAs although significantly lower than when other solutions were present, were still fairly high. This could be due to several factors. First, Bartzat 2000 suggests that these chemicals can only be detected at close range. Therefore, the response to the PAs would need to be learned and thus high visits could be due to a slow learning response or simply initial visits by different butterflies. If this study were to be con ducted again, the breakdown of when visits occurred during the hour intervals should be recorded. Second, because Heliconiine butterflies are very long lived and this experiment was done in an unnatural environment where most of the subjects were raised i n captivity and have been visiting the same flowers for months, they most likely have a fairly ingrained routine and
preference; this also could lead to a delayed response. The need for field tests on this subject is great. However, findings from the vis itation duration experiments help to explain how PAs could deter but still result in a high frequency of visits. For flowers with PAs, the duration of visits greatly decreased. This trend was not observed in the manipulated flower groups for any of the o ther treatments sucrose or empty. Although the empty treatment yielded significant changes in the duration of visits for certain groups of flowers, the changes were not observed for the treated groups. For this phase of the experiment, the biggest sour ce of change between the sucrose and empty treatments was observed for the S. frantzii flowers which were not manipulated. After the L. camara flowers were made empty, duration of visits to the S. frantzii switched from primarily long, to medium visits. Visit duration did not change significantly for any other group during this period. Because this part of the experiment was only conducted for one day, this observation could simply be an artifact of small sample size. Some of the observations for durat ion of visits to flowers containing PAs could be biased towards being deceptively high. Casual observations noted that when a butterfly would visit one of these flowers, they would test each flower on the inflorescence before flying away. So, even though they were not drinking the nectar visits could still be recorded as long if the inflorescence was large. Therefore, this study is conservative in its findings. Better methods need to be developed to better describe the exact response of butterflies to solutions containing PAs. Different flowers without so many located close together on an inflorescence could diminish this bias as well. Despite certain drawbacks, the data indicate the when PAs are present visits to flowers decrease in number and visi tation duration. Therefore, it is fair to say that pyrrolizidine alkaloids deter generalist Heliconiine butterflies. Further studies should be conducted to examine the effectiveness of PAs to deter other species of generalist pollinators. Unfortunately due to small numbers of individuals and time of year, the portion of this study aimed to determine if PAs attract the specialist Ithomiine butterflies had to be discarded. However, this idea is well supported in the literature Masters 1968, Pilske 1975, Boppre 1978, Brown 1984, Bartzat 2000. A study in Monteverde concerning this subject needs to be conducted from May to August in order to observe these butterflies before migration occurs Goode 1999. Color choice could not be established from experime nts 1, 3, 4 and 5. This is most likely due to low visits to these flowers small sample size. High frequency of visits to Lantana camara which is a primarily orange morph could suggest that this is their color preference. These results again could be biased because of the location of the study: the butterflies are long lived and have already established preferred flowers. However, their hesitancy to switch to these offered flowers suggest that although these butterflies are believed to be generalists , they form strong preferences to certain types of flowers and may be more efficient pollinators than other generalist species. This suggestion is purely hypothetical and studies are needed to examine this possibility. The data also suggest that pyrroliz idine alkaloids act to deter nectar robbers. For all three types of ants, the average number of individuals found at each solution was lowest for the solution containing PAs. For the BB ants PA pools had significantly fewer visits than any of the other s olutions containing plant products. The fact that no
significant differences were found between the PA solution and the other solutions with plant matter for the other ant types could be a result of many factors. First, both the MRB and TR ants were on m any fewer plots and thus the sample sizes were small for the two groups. These species of smaller ants may have been on fewer plots because more individuals are able to visit a single plot at one time. Similar numbers of individuals of the different spec ies could have been observed, however, results were less significant for the smaller sized species because statistical significance was based upon number of plots, not number of individuals. Second, perhaps the other two plant types used to make the contr ol solutions Lantana sp. and A. pinchensis have different chemicals that act as deterrents as well. However, this is highly unlikely since significant differences were not found between the sucrose solution and these plant solutions p<0.05. Further st udies are needed to examine how different insects and possible nectar robbers respond to PAs. It can be concluded then that the function of pyrrolizidine alkaloids in nectar is not only to attract a specialist pollinator but also to deter generalist pol linators and nectar thieves. Just as secondary compounds in vegetative plant parts act to deter generalist herbivores and reduce vegetative losses, toxic nectars may act to protect reproductive parts and offerings. Toxic nectars appear to be adaptive and help to increase reproductive fitness in plants that produce them. Therefore, plant species that produce PAs in their nectar are helping to increase their reproductive fitness by promoting pollinator constancy while inhibiting energetic reproductive loss es due to insufficient pollination and nectar thievery. ACKNOWLEDGMENTS Alan Masters &did you ever know that you re my hero?!?! Thank you so much for the countless hours you allocated to encourage my growth as a student of biology and life. Your g uidance, patients, brilliance, understanding and open mindedness carried me through this experience and helped me gain insights to many aspects of myself and the world. I would also like to thank Maria and Ollie for the help with stats, materials and for comforting me through the crazy, stressful times and for always providing a good laugh. Of course, many thanks to Karen for helping me keep a peace of mind, for the willingness to open your arms and doors, and for guidance and reassurance in general and the coffee and cake Ã°J . Javier, you silly tico &thank you for helping me keep a smile on my face, for the knowledge you ve provided me throughout the semester and for helping write my resumen. Thanks to Jim Wolfe, Julio, and the rest of the butterfly garde n gang for the open doors, laughs, and support. Thank you Lauren Bennett for helping me with graphs, tables and to find sources when time got crunched, as well as, for your strength and support &I might have gone insane without your positive energy. Finall y, many thanks to Kevin Loope for not only acting as my alarm clock during late night naps on the floor but also as a constant source of encouragement, comfort and knowledge. LITERATURE CITED Appanah, S. 1990. Plant pollinator interactions in Malaysi an rain forests. In K.S. Bawa and M. Hadley Ed.. Reproductive Ecology of Tropical Forest Plants, pp. 85 101. Unesco and The Parthenon Publishing Group Limited, Carnforth, UK. Baker, H.G., and I. Baker. 1975. Studies of nectar constitution a nd pollinatior plant coevolution. In L.E. Gilbert and P.H. Raven Eds.. Coevolution of animals and plants, pp.100 140. University of Texas Press, Austin, Texas.
Baker, H.G. 1978. Chemical aspects of the pollination of woody plants in the tropics. Pages 57 82 in P.B. Tomlinson and M. Zimmerman, eds. Tropical trees as living systems. Cambridge University Press, New York. In Rhoades, D.F., Bergdahl, J.C. 1981. The Significance of Toxic Nectar. The American Naturalist 175: 798 803. Baker, H .G., and Baker, I. 1982. Chemical constituent of nectar in relation to pollination mechanisms and phylogeny, pp. 131 171 in M.H. Nitecki ed.. Biochemical Aspects of Evolutionary Biology. University of Chicago Press, Chicago, Illinois. Bartzat, A. 2000. Ithomiine Butterfly Attraction to Ageratum rugosa. Tropical Ecology and Conservation. CIEE. Bawa, K.S. 1990. Plant pollinator interactions in tropica rain forests. Annual Review of Ecological Systems 21: 399 422. Boppre, M. 1978. Chem ical communication, plant relationships, and mimicry in the evolution of danaid butterflies. Entomol. Exp. Appl. 24:264 77 In Nishida, R. 2002. Sequestration of defensive substances from plants by Lepidoptera. Annu. Rev. Entomol. 47:57 92. Brown, K. S., Jr. 1985. Chemical ecology of dehydropyrrolizidine alkaloids in adult Ithomiinae Lepidoptera, Nymphalidae. Rev. Brasil. Biol. 44:435 460 In Masters, A.R. 1991. Dual role of pyrrolizidine alkaloids in nectar. Journal of Chemical Ecology 17:1 95 206. Brown, K.S., Jr., et al. 1991. Aposymatic insects on toxic host plants:coveolution, colonization, and chemical emancipation. In P.W. Price and T.M. Lewinsohn Eds.. Plant animal interactions, pp.375 402. John Wiley and Sons, Inc. New York. DeVries, P.J. 1987. The butterflies of costa rica and their natural history. Volume I: Papilionidae, Pieridae, Nymphalidae. Princeton University Press. Princeton, New Jersey. Endress, P.K. 1994. Diversity and evolutionary biology of tropical flow ers, pp.122 188. Cambridge University Press. New York, New York. Ehrlich, P.R. and Raven, P.H. 1964. Butterflies and plants: A study in coevolution. Evolution 18:586 608 Goode, M.R. 1999. An Introduction to Costa Rican Butterflies. 1a. ed. San J ose. Ghazoul, J. Can floral repellents pre empt potential ant plant conflicts? Ecology Letters 4: 295. Haber, W. 1978. Evolutionary ecology of tropical mimetic butterflies Lepidoptera: Ithomiinae. Ph.D. dissertation, Univ. of Minnesota. In DeVri es, P.J. 1987. The butterflies of costa rica and their natural history. Volume I: Papilionidae, Pieridae, Nymphalidae. Princeton University Press. Princeton, New Jersey. Hannan, G.L. 1981. Flower color polymorphism and pollination Biology of Play tstemon californicus Benth. Papaveraceae. Anerican Journal of Botany, Botanical Society of America. Janzen, D.H. 1977. Why don t ants visit flowers? Biotropica 9:252. Janzen D.H. 1981. Patterns of herbivory in a tropical deciduous forest. Biotro pica. Kearns, C.A. and Inouye, D.W. Techniques for Pollination Biologists, pp.200 215. University Press of Colorado. Niwot, Colorado. 1993.
Landolt, P.J., Lenczewski, B. 1993. Lack of evidence for the toxic nectar hypothesis: A plant alkaloid did n ot deter nectar feeding by Lepidoptera. Florida Entomologist 76:556 566. Levin, D.A. 1975. Pest pressure and recombination systems in plants. American Naturalist 109:437 51 In Bawa, K.S. 1990. Plant pollinator interactions in tropical rain fores ts. Annual Review of Ecological Systems 21: 399 422. Pilske, T.E. 1975. Attraction of Lepidoptera to plants containing pyrrolizidine alkaloids. Environ. Entomol. 4:474 479. Ray, T., and Andrews, C.C. 1980. Ant butterflies: butterflies that follow ar my ants to feed on antbird droppings. Science 210:1147 1148. In DeVries, P.J. 1987. The butterflies of Costa Rica and their natural history. Volume I: Papilionidae, Pieridae, Nymphalidae. Princeton University Press. Princeton, New Jersey. Master s, A.R. 1990. Pyrrolizidine alkaloids in artificial nectar protect adult ithomiine butterflies from a spider predator. Biotropica 22:298 304. Masters, A.R. 1991. Dual role of pyrrolizidine alkaloids in nectar. Journal of Chemical Ecology 17:195 206. M asters, J.H. 1968. Collecting Ithomiidae with heliotrope. J. Lepidop. Soc. 22:108 109 In Masters, A.R. 1991. Dual role of pyrrolizidine alkaloids in nectar. Journal of Chemical Ecology 17:195 206 Morris, A. 2005. Functional Differences Among Colo r Morphs of Impatiens walleriana Balsaminaceae. Tropical Ecology and Conservation. CIEE. Raven, P.H., Evert, R.E., and Eichorn, S.E. 1986. Biology of Plants, pp.590 620. Worth Publishers. New York, New York. Rhoades, D.F., Bergdahl, J.C. 1981. The Significance of Toxic Nectar. The American Naturalist 175: 798 803. Simpson, B.B., and J.L. Neff. 1983. Evolution of diversity of floral rewards. In C.E. Jones and R.J. Little Eds.. Handbook of experimental pollination biology, pp.142 15 9. Van Nostrand Reinhold, New York, New York. Stephenson, A.G. 1982. Iridoid glycosides in the nectar of Catalpa speciosa are unpalatable to nectar thieves. Journal of Chemical ecology 8:1025 34.