Resource Partitioning Between Trigona fulviventris and Scaptotrigona mexicana With Overlapping Flight Ranges Michelle H. Averbeck Department of Biological Sciences, Ecology, Behavior and Evolution Program, University of California, San Diego _____________________________________________________________________________________ ABSTRACT The theory of resource partitioning predicts that congeneric species are allowed to coexist by a division of the available resources. This study looks at the pos sibility of resource partitioning between Trigona fulviventris and Scaptotrigona mexicana on fine temporal scales in Cloud Forest habitat, a region in which stingless bee pollen diets has not been extensively studied. Workers from one T. fulviventris and o ne S. mexicana nest, located 200 m apart, were studied synchronously over six days. Capture mark and release experiments verified that the foraging areas of these two nests indeed overlap. Of the 14 morphotypes of pollen brought in by S. mexicana and the 1 6 morphotypes brought in by T. fulviventris seven morphotypes were found in common between the two species Srenson index = 0.3. RESUMEN La teora de la divisin de los recursos predice que los especies similares son permitidos coexistir para una divisin de los recursos disponibles. Este estudio ve a la posibilidad de la divisin de los recursos entre de Trigona fulviventris y Scaptotrigona mexicana a bajo plazo en el hbitat del bosque de los nubes, una regin en que las alimentaciones de polen de abejas sin picaduras no hubo estudiado extensamente. Trabajadores de T. fulviventris y de S. mexicana con los nidos 200 metros afuera de juntos, estudiaba a el mismo tiempo por seis das. Experimentos de capturar marcar y soltar confirmaron que las re as de vuelo de esos dos nidos traslapan. De los 14 morfotipos de polen trajo al nido de T. fulviventris y los 16 de S. mexicana siete morfotipos compartan entre de los dos especies ndice de Srenson = 0.3. INTRODUCTION Stingless bees Apidae: Melipo ninae are the predominant bees of the tropics Hanson 1995. In Guanacaste Province, Costa Rica they comprise nearly 30 percent of the insect community Johnson and Hubbell 1974. In Central America, eusocial tropical stingless bee communities exist in mi xed species and conspecific communities Johnson & Hubbell 1978. Female workers collect pollen to feed to larvae Hanson 1995. The high diversity of pollen producing plants in the tropics may favor generalist consumers if the resources are temporally and spatially heterogeneous Heinrich 1976. Because stingless bees are generalists they may be afforded to facul atively switch to differing resources. This may decrease competition, allowing similar species to coexist. Generalist pollen gathering in bees may be an adaptation to deal with the toxicity of pollen of some flowering species Roubik 1989. Stingless bees are usu ally generalists when examined over time scales of months Hubbell and Johnson 1977. Additionally, by shifting activity periods to diminish competition, stingless bees may temporally partition resources De Bruijn 1997. When foraging for pollen, stingle ss bees visit the foraging site, pack pollen grains on their hind legs and return to the nest. Foraging occurs in an area surrounding the nest Hubbell & Johnson 1977. The size of the foraging area is limited by maximum flight range, which is about 1.2 km for stingless
bees in Guanacaste Province, Costa Rica Roubik 1989. In Brazil, a smaller species of Trigona was trained to fly to an attractive scent, then marked, and their return to the nest was observed and recorded. The greatest distance at which ind ividuals were flying to and returning from the sucrose source was 500 m Juliana Rangel, pers. comm. 2003. In Monteverde, Costa Rica, Trigona fulviventris and Scaptotrigona mexicana nest 200 m within each other, and therefore have access to similar resou rces. Stingless bees have exhibited an overlapping of the floral species they choose to visit in long term studies Johnson and Hubbell 1974. However, two species with overlapping foraging areas and similar foraging strategies, may compete for resources o n finer temporal scales. Others have looked at interspecific resource partitioning in Trigona Eltz et al. 2001. Interspecific aggressive foraging behavior was observed by Johnson and Hubbell 1974 suggesting competition. Another study done in Malaysia s uggested resource partitioning occurred through preferences in foraging strata Nagamitsu et al. 1998. If two similar stingless bee species were competing for a limiting resource, resource partitioning would lessen the eliminating effects sometimes cause d by competition. This study looks first to see if T. fulviventris and S. mexicana exhibit an overlapping foraging area and second to see if they are partitioning their pollen resources on small temporal scales or usurping the same resources. MATERIALS AN D METHODS Study Site The study took place at 1450 m on the Pacific slope of the Cordillera de Tilarn at the Estacin Biolgica, Monteverde, Costa Rica from April 9 May 5, 2003. Two nests were chosen that were 200 m apart and located in open areas within 100 m from lower montane cloud forest Nadkarni and Wheelwright 2000. Flight Range Experiment Sixty five Trigona fulviventris individuals were collected using hand held aspirators or pheromone traps see Pollen Collection, marked and released at 100m the midpoint between the two colonies and 500 m from the nest. Individuals were marked with paint pens on the thorax following fifteen to twenty minutes of refrigeration. Four mark and release events took place, resulting in the release of 65 individuals at each distance. Nests were monitored for one hour, 24 hours and 48 hours after the release, and returning individuals were collected and terminated to avoid recounting individuals. Pollen Collection Bees were always collected between 0900 and 1300 hours. If individuals carrying pollen were collected from one nest, collections were made from the other nest within 24 hours. These consecutive days of sampling for both nests are referr ed to as paired data days. The nests were tested in paired data days three times, resulting in six days of pollen collection total. A seventh day of data collection was used to sample exhaustively for pollen morphotypes and was not included in pair wise co mparisons. For every day of data collection, 30 bees were collected using a hand held aspirator. Individuals were also collected using a pheromone trap. This involved placing a small plastic funnel over the colony entrance with a mesh bag attached to the f unnel stem. Once an individual wandered into the bag, it was perturbed to release an alarm pheromone, hence attracting other individuals into the mesh bag. The bag was then left unattended and observed in 15 minute intervals until enough bees had been col lected. The bees were next cooled for 15 20 minutes in a refrigerator. While the bees were experiencing
reduced mobility due to being cold, individuals carrying pollen loads were found, placed in separate glass vials and subsequently frozen. The remaining bees regained mobility within ten minutes and were released 50 m from the nest. Pollen Analysis Pollen from every individual collected was identified to morphotype. Frozen bees were scraped for pollen samples, and pollen was placed on a glass slide and s wirled in water to separate pollen grains. Samples were viewed under a monocular compound scope with 40 X power. Each sample was catalogued based on size, shape, texture, and color. A library of pollen morphotypes was created with hand drawings and detaile d descriptions, and each morphotype was assigned a corresponding letter for reference. Morphotype per sampled individual was recorded and Shannon Weiner index was used to find species richness, and morphotype individual curves were constructed. A Srenson index was used to test for the overall degree of morphotype overlap between T. fulviventris and S. mexicana and a Chi squared test was used to compare the differences between species within paired data days. RESULTS Flight distance and return frequency Trigona fulviventris returned to the nest from both 100m and 500 m. Nineteen of 65 marked Trigona fulviventris released at 100 m from the nest returned to the nest within 48 hours. Three of 65 marked T. fulviventris released at 500 m from the nest returne d to the nest within 48 hours. Bees released at 100 m returned more often than those released at 500 m Chi squared test: X 2 = 14.0007, P = 0.0002 Table 1. ________________________________________________________________________ TABLE 1. Distances from nest at which Trigona fulviventris were released and the frequency of the recorded returns in marked individuals. ________________________________________________________________________ Released from 100m Released from 500 m Number of bees released 65 65 Number of bees observed returning 19 3 Return frequency 29.2% 4.6% OVERALL POLLEN RICHNESS AND OVERLAP In three sampling periods resulting in 70 total individuals Scaptotrigona mexicana brought in 14 morphotypes, mostly of types B and I, and T. fulviventris brought in 16 of mostly H and K Figure 1. Pollen morphotype richness versus individuals sampled curves were similar for both nests Figure 2. Shannon Weiner indices for richness of total pollen loads and for paired data days were similar m odified t tests, P > 0.05. Twenty three morphotypes were catalogued, and the two coloniesÂ€ pollen load diversity overlapped by seven morphotypes Srenson index = 0.3 Figure 3. Within data sets, differences in morphospecies were more abundant than simi larities Figure 4. The Srenson index was found to be 0.3, reflecting the seven morphotypes that were in common between the two species.
POLLEN RICHNESS AND OVERLAP FOR PAIRED DATA DAYS Within the three paired data days, frequency of morphotypes was f ound to be random for the first paired data days. For paired data days two and three, pollen morphotypes brought in by T. fulviventris and S. mexicana were found to be significantly different see Second and Third paired data days. First paired data days A total of nine morphotypes of pollen were found on T. fulviventris and S. mexicana eight on T. fulviventris three exclusively and six on S. mexicana one exclusively. Five morphotypes were in common between both species. Significance of frequency of workers carrying the nine morphotypes of pollen was random Chi squared test; X 2 = 10.786, P > 0.05, n T. fulviventris = 16, n S. mexicana = 20 Figure 3a. Second paired data days Fourteen morphotypes of pollen were recorded for both species. All excep t for one morphotype of pollen type K were found exclusive to either T. fulviventris or S. mexicana T. fulviventris brought in seven morphotypes, and S. mexicana brought in eight. The frequency of the workers carrying the 14 morphotypes of pollen was found to be significantly different between T. fulviventris and S. mexicana Chi squared test; X 2 = 49.746, P < 0.0001, n T. fulviventris = 26, n S. mexicana = 29 Figure 3b. Third paired data days T. fulviventris and S. mexicana brought in eight morphotypes. Again, all except for one morphotype of pollen type H were found to be exclusive. Five morphotypes were found on T. fulviventris and S. mexicana brought in four. The frequency of the workers carrying the 14 morphotypes of pollen wa s found to be significantly different between T. fulviventris and S. mexicana Chi squared test; X 2 = 27.598, P = 0.003, n T. fulviventris = 14, n S. mexicana = 21 Figure 3c.
________________________________________________________________________ FIGURE 1 Black bars represent pollen morphotypes found on individuals of Scaptotrigona mexicana Gray bars represent pollen morphotypes found on individuals of Trigona fulviventris _______________________________________________________________________ ________________________________________________________________________ FIGURE 2. Richness of pollen morphotypes per individual sampled. Black line is species richness curve for Scaptotrigona mexicana. Gray line is species richness curve for Trigona fulviventris ________________________________________________________________________
________________________________________________________________________ FIGURE 3. Each bar represents a morphotype of pollen brought in by either nest during one of paired data days. Black bars represent percent of pollen morphotype found on individuals of Scaptotrigona mexicana Gray bars represent percent of pollen morphotype found o n individuals of Trigona fulviventris A solid black or solid gray bar means that morphotype was only found on that species. a Is the first paired data days, b second paired data days and c third paired data days ____________________________________ ____________________________________
DISCUSSION Flight distance and return frequency Trigona fulviventris has a foraging area that overlaps with S. mexicanaÂ€s nest location. Theoretical flight distributions can be created with a beeÂ€s maximum flight range Hanson 1995. Three T. fulviventris individuals returned from a distance of 500 m from the nest, and their ability to return to the nest increased with decreasing distance with 19 individuals returning from a distance of 100m. A flight range experiment in Brazil also showed Trigona spÂ€s maxi mum flight range to be 500m Juliana Rangel, pers. comm. 2003. Other extensive flight experiments show the majority of foraging takes place one third to one half of the foraging range Hanson 1995. Therefore, the T. fulviventris workers may spend most of their time foraging 250 300m from the nest with a maximum flight range of 500m. These results are interesting because the two nests included in the study were located only 200 m apart, which puts T. fulviventris well into S. mexicanaÂ€s home range. Althoug h this study did not take foraging direction into account, the results point to an overlapping home range between Trigona black and Trigona yellow, and if each species forages in a different direction they may avoid competition through spatial differentiat ion. Pollen richness and overlap Trigona fulviventris and Scaptotrigona mexicana brought in 23 different morphotypes of pollen overall, and seven of 19 morphotypes were shared between them during paired data days. These seven morphotypes in common reveal overlap in resource foraging. However, within the 24 hour time frame of the paired data days, two of the three paired days showed a non random frequency of workings bring in pollen, and the morphotypes brought in by both species were significantly differe nt Figure 3b & 3c. This suggests that on finer time frames T. fulviventris and S. mexicana may avoid resource overlap. For the second and third paired data days morphotype overlap was low with only one morphotype in common in each case. However, resourc e overlap was higher for the first paired data day with five morphotypes in common Figure 3a. Flower phenology may have resulted in a temporary abundance of pollen resources, and in the absence of a limiting resource the two species would not be benefite d by the decrease in competition associated with resource partitioning. Other studies have also shown that although there is substantial overlap in the type of flowers stingless bees visit in the long run, resource partitioning indeed exists between eusoci al tropical bees on finer temporal scales Eltz et al. 2001, Sommeijer et al. 1983, Nagamitsu et al. 1999. Evidence for resource partitioning This study suggests pollen resources are being partitioned as a way for similar stingless bee species to coexis t. Long term studies have shown resource overlap, but my results show that resources are partitioned on fine time frames. Many speculations can be made about the selection force that resulted in this pattern of resource partitioning. Past studies have sugg ested spatial or temporal differences in flower search behavior or foraging strategy may lead to a division of resources. Eltz et al. 2001 Johnson and Hubbell 1974 observed aggressiveness as evidence of competition between species of stingless bees in Costa Rica. The aggressiveness of one species over another determined which species has access to the most coveted floral resources. According to the original resource partitioning theory, it is indeed the partitioning of resources that allows species to c oexist and prohibits the competitive elimination of all but one species MacArthur 1958.
Other hypotheses for differential resource use between bee species have included that different morphological structures e.g. proboscis length allow some species t o obtain resources from those flowering species with complementary floral structures Johnson R. 1986. However, T. fulviventris and S. mexicana have similar body sizes, and the purpose of this study was to verify resource partitioning, not the selective forces behind it. Due to the substantial evidence pointing towards the existence of resource partitioning in stingless tropical bees, the mechanisms of resource partitioning among stingless bees should be examined. While long term studies have found that there in resource o verlap, this study suggests in the short term, they are partitioning their pollen resources. Because T. fulviventris and S. mexicana are visiting different flowers on fine temporal scales Figure 3b & 3c, they are allowed to coexist. With the ex perimental ly derived overlapping foraging range and the mapped and recorded differences in the pollen diets of T. fulviventris and S. mexicana it can be concluded that interspecific resource partitioning is maintaining these populations by reducing competition. ACK NOWLEDGEMENTS I would like to thank Karen Masters for her support and guidance and her ever encouraging advice of ÂJust do it!Â‚ I also thank Chrissi Murphy, Frank Joyce, Juliana Rangel, Tracy Rogers and Ramon Amores Campos for their help in formulating th is study. Special appreciation goes to Paul Hanson for identifying my stingless bees. LITERATURE CITED De Bruijn, L.L.M. and M.J. Sommeijer. 1997. Colony foraging in different species of stingless bees Apidae, Meliponinae and the regulation of individu al nectar foraging. Insects soc. 44: 35 47. Eltz, T.A., C.A. Brhl, S. van de Kaars, V.K. Chey and K.E. Linsenmair. 2001. Pollen Foraging and resource partitioning of stingless bees in relation to flowering dynamics in a Southeast Asian tropical rainforest Insects soc. 48: 273 279. Hanson, P.E. and I.D. Gould. 1995. The Hymenoptera of Costa Rica, Oxford University Press, Oxford Heinrich, B. 1976. Resource partitioning among some eusocial insects: Bumblebees. Ecology 57: 874 889. Hubbell, S.P. and L.K. John son. 1978. Comparative foraging behavior of six stingless bees exploiting a standardized resource. Ecology 59: 1123 1136. _______, and L.K. Johnson. 1977. Competition and nest spacing in a tropical stingless bee community. Ecology 58: 949 963. Nagamitsu, T ., K. Momose, T. Inoue and D.W. Roubik. 1999. Preference in flower visits and partitioning in pollen diets of stingless bees in an Asian tropical rainforest. Res. Popul. Ecol. 41: 195 Âƒ 202. _______, K. Momose, T. Inoue. 1998. Interspecific morphological v ariation in stingless bees Hymenoptera: Apidae, Meliponinae associated with floral shape and location in an Asian tropical rainforest. Entomological Science 1: 189 194. Sommeijer, M.J., G.a. de Rooy, W. Punt, and L.L.M. De Bruijn. 1983. A comparative stu dy of foraging behavior and pollen resources of various stingless bees Hym., Melinponinae and honeybees Hym., Apinae in Trinidad, West Indies. Apidologie 14: 205 224. Johnson, L.K. and S.P. Hubbell. 1974. Aggression and competition among stingless bees : Field studies. Ecology 55: 120 127. Johnson, R.A. 1986. Intraspecific resource partitioning in the bumble bees Bombus ternarius and B. Pennsylvanicus Ecology 67: 133 138. Mac Arthur, R.A. 1958. Population ecology of some warblers of Northeastern coniferous forests. Ecology 39: 599 619. In Foundations of Ecology. Edited by L.A. Real and J.H. Brown. Foundations of Ecology. 1991. The University of Chicago, Chicago. Nadkarni, N.M. and Wheelwright N.T. Ed. 2000. Monteverde: Ecology and conservation o f a tropical cloud forest. Oxford University Press, New York, New York. Roubik, D.W. 1989. Ecology and Natural History of Tropical Bees. Cambridge University Press, New York.
Tree Size and Habitat Effects on Stream Gall Abundance in Conostegia oerstediana (Melastomataceae) Priya Shashidharan Department of Earth and Environmental Sciences, George Washington University _____________________________________________________________________________________ ABSTRACT The moth Mompha sp. (Coleophoridae, Lepidoptera) is known to induce a stem gall on the tree Conostegia oerstediana (Melastomataceae). There is little known abo ut the distribution and abundance of galls. This study tested the difference in stem gall abundance between varying tree sizes and between two different habitats pasture and secondary forest. Trees from each habitat were sampled and measured for diameter at breast height (DBH), height, number of branches, and number of galls. A significant difference was found between pasture and forested areas (unpaired t test, p < 0.0001), with pasture trees having more galls. No relation was found relating tree size (DB H, height, number of branches) to gall abundance. I conclude from these results that tree size is not directly related to stem gall abundance. Instead, differences in habitats, such as predator and parasite abundance and host density, may be important fact ors that influence gall abundance. RESUMEN La polilla Mompha sp. (Coleophoridae, Lepidoptera) induce una agalla de tallo en el rbol Conostegia oerstediana (Melastomataceae ). Poco es conocido de la distribucin y abundancia de agallas. En este estudio se examin la diferencia en la abundancia de agallas de tallo entre rboles de tamaos diferentes y entre dos ambientes el pasto y el bosque secundario. Arboles de cada hbita t fueron medidos por DBH, altura, cantidad de ramas y cantidad de agallas. Una diferencia significante fue encontrado entre el pasto y el bosque (prueba de t unpariada, p < 0.0001), con el pasto poseyendo ms agallas. No se encontr una relacin entre el t amao del rbol (DBH, altura, cantidad de ramas) y la abundancia de agallas. De estos resultos concluyo que el tamao de rbol no es relacionado directamente a la abundancia de agallas de tallo. En vez de eso, diferencias entre hbitats, como la abundancia de predators y parsitos y la densidad de rboles, podran ser factores importantes que influyen la abundancia de agallas.
INTRODUCTION A great number of plant taxa harbor growth abnormalities known as galls. A gall is a mass of swollen plant tissue t hat develops in response to the parasitic attack of certain species of insects, bacteria, fungi, spiders and mites (Mani, 1992). These gall formers live inside the structure, using it for nutrition and protection against parasites, predators, disease, and harsh environmental conditions (Price et al. 1987). Gall tissue is an excellent food source, being highly nutritive, even more so than the rest of the plant. Galls also contain a decreased amount of chemical defenses (Price et al. 1987). Effects on the pla nt are quite detrimental. There is evidence that gall production prevents pollen and seed development (Graham, 1995) and may even cause shoot and branch death (Price et al, 1987). Galls occur on all plant organs, from roots to ovaries to leaves, and have a wide range of shapes and sizes (Mani, 1992). The appearance of a gall and where it is located on the plant is determined by the gall forming species. This host specific relationship is obligatory for the development of the galler. The formation of insect induced galls is initiated by oviposition or the feeding of the first instar larva. The mechanism of gall formation is not well known, but it has been hypothesized that either the mother, larva, or both inject or secrete a chemical that redirects and prom otes growth of undifferentiated plant tissue (Borror et al. 1989; Evans, 1984; Hogue, 1993). Five orders of insects contain gall making species; Diptera, Homoptera, Hymenoptera, Coleoptera, and Lepidoptera. Of the approximately 2000 species of gallers in North America, 1500 are gall wasps (Hymenoptera, Cynipidae) or gall midges (Diptera, Cecidomyidae) (Evans, 1984). Lepidoptera are not known to be common gallers, especially in Costa Rica, where galling insects are generally uncommon. However, Lepidopterans of the genus Mompha (Coloephoridae, Momphinae) are observed to make at least four different gall structures in the neotropics; three on Cuphea (Lythraceae) and another on Conostegia oerstediana (Melastomataceae). Mompha are microlepidopterans with long, n arrow wings. The larvae are herbivores that typically feed on leaves, buds, and flowers (Graham, 1995). Conostegia oerstediana is a dominant secondary forest tree species in Monteverde, Costa Rica that is also typical in old pastures and forest edges (Hab er et al. 2000). This tree has been observed to be infested with galls on both its leaves and stems, the spherical stem galls being induced by Mompha sp In this study I will examine the relationship between C. oerstediana and Mompha sp to determine the e ffect of tree size and habitat on the abundance of galls. There are two hypotheses that explain why galls may occur more often in one place than another. One is the spatial heterogeneity hypotheses (Akimoto, 1994), which predicts that gallers would choose plants with larger leaves and shoots, as this shows higher nutrition availability. The other is the synchronization hypotheses (Akimoto, 1994), which predicts that gallers choose and are most successful when they attack plants at a certain time, specifica lly bud burst. I hypothesize that there are many factors that vary between habitats that are likely to have an effect on gall abundance, such as availability of nutrients, presence of parasites and predators, and host density. I also hypothesize that tree size and number of galls per tree will have a positive relation. A larger tree provides a larger area for gallers to oviposit, and therefore they should be
directly related. The purpose of this study is to determine whether tree size and habitat effects be tween pasture and secondary forest change the abundance of the C. oerstediana stem gall. MATERIALS AND METHODS Sample Site elevations ranging from 1400 1700m during April 10, 2003 May 10, 2003. Measurements and collections of C. oerstediana were made at two secondary forest sites and two pasture sites. The two secondary forest sites were located in the surrounding forest of the Estacin Biolgica de Monteverde (EBM). The pasture sites were located north of the EBM and at the farm of Federico Muoz in Las Nubes. Data collection Secondary forest trees were sampled by using every other C. oerstediana encountered along a 50m transect. In pasture areas, each isolated C. oerstediana tree encountered was sampled. Diameter at breast height (DBH) and tree height were measured. Number of branches was counted, defining a branch as the extension of the tree that usually bore the gall (typically the quaternary branch). Number of galls per branch was counted with the aid of binoculars. In secondary forest distance to nearest neighbor was also measured. A total sample size of 55 trees, 30 from secondary growth and 25 from pasture, was obtained. A maximum of 20 galls that were lower than six meters was collected per tree, using a tree pruner if necessary, and labeled by tree number. In the lab, the gall was cut open using a pocketknife and examined for presence of larvae, pupae, and adults. If larvae were found, the width of the base of the head was measured with a dissecting scope and an ocular micrometer to determine larval instar. Statistical Analysis Unpaired t tests between DBH, tree height, number of branches, and gall abundance in pastures and secondary growth were performed. A simple regression analysis related DBH, tree height, and number of branches (and for secondary growth distance t o nearest neighbor) to gall abundance for each habitat. Additionally, larval head sizes were plotted in a histogram. The histogram was divided into seven intervals to show the typical seven instars of a moth larva. RESULTS In secondary growth a total of 30 trees were sampled. The mean DBH was 11.7 11.6 cm. The mean tree height was 6.6 3.2m. The mean number of branches was 381 831.9. Pasture trees were generally larger than secondary forest ones and had a small range of sizes. The mean DBH was 53.1 30.8cm. The mean tree height was 8.5 2.8m.
The mean number of branches was 2133 1699.5. Number of galls per tree was also higher in pasture. Pasture trees had a mean of 115. 3 97.6 galls per tree whereas secondary growth trees had a mean of 4.1 4 .9 galls per tree (Table 1). A significantly greater abundance of galls was found in pasture trees than in secondary growth trees (unpaired t test, p < 0.0001) (Fig. 1). Significant differences were also found between DBH (unpaired t test, p < 0.0001) (Fi g. 2), tree height (unpaired t test, p = 0.0282) (Fig. 3), and number of branches (unpaired t test, p < 0.0001) (Fig. 4) between the two habitats, with pasture trees being bigger in all cases. There was no relation between number of galls in the pasture v ersus DBH (simple regression, p = 0.595, r 2 = 0.013) (Fig. 5), tree height (simple regression, p = 0.716, r 2 =0.006) (Fig. 6), or number of branches (simple regression, p = 0.392, r 2 = 0.032) (Fig. 7). A positive relationship was found between DBH versus n umber of galls in secondary growth (simple regression, p = 0.004, r 2 = 0.265) (Fig. 8) and number of branches versus number of galls in secondary growth (simple regression, p = 0.008, r 2 =0.227) (Fig. 10), but there was no relationship between height (simp le regression, p = 0.269, r 2 = 0,043) (Fig. 9) or nearest neighbor (simple regression, p = 0.094, r 2 = 0.097) (Fig. 11). Head sizes of larvae found in pasture were also plotted in a histogram, which revealed that many of the larvae sampled (77 out of 174) were in their sixth instar (head size between 12.3 14.3 mm) (Fig. 12). DISCUSSION This study showed that there was a significant difference in tree size between pasture and secondary growth sites. Pasture trees were found to be significantly larger in terms of DBH (unpaired t test, p < 0.0001) (Fig. 2), tree height (unpaired t test, p = 0.0282) (Fig. 3), and number of branches (unpaired t test, p < 0.0001) (Fig. 4). Because C. oerstediana is a dominant tree in the secondary forest of Monteverde and oft en develops nearly single species stands (Haber et al. 2000), there is intense competition for resources, such as light, space, and nutrients. This strong intraspecific competition in secondary growth may be what causes the trees in that habitat to grow sk innier, shorter, and with fewer branches than pasture trees (Fig. 2, 3, 4). Pasture trees do not have to live in such conditions of high competition because they are isolated, and therefore can utilize resources to a greater extent, consequently forming la rger, more robust individuals (Aldrich & Hamrick, 1998). There was no significant relation found between number of galls per tree and tree size in terms of DBH (simple regression, p = 0.595, r 2 = 0.013), tree height (simple regression, p = 0.716, r 2 = 0.00 6), and number of branches (simple regression, p = 0.392, r 2 = 0.032) in pasture trees (Fig. 5, 6, 7). Larger trees do not necessarily have more galls because there may be other factors that influence host selection by gall formers. One, as suggested by th e synchronization hypothesis (Akimoto, 1994), may be the developmental stage of the plant host organ. Other factors related to habitat differences ay also influence gall abundance. A significant difference was found in stem gall abundance between pasture and secondary growth sites (unpaired t test, p < 0.0001) (Fig. 1). Pasture trees were found to have a greater number of galls per tree than secondary forest trees (Table 1). Abiotic conditions in pasture sites are notably different than in secondary growth sites. Pasture sites are more exposed to wind and sunlight, have higher temperatures, and have lower
humidity. These desiccating conditions are inhospitable to many species of arthropods. A study done by Lori Olson (1994) in Monteverde, Costa Rica found that species diversity of arthropods is lower in open habitats than in forested ones. Gall making species, however, should not be affected by these abiotic conditions, as gall formation begins at the time oviposition, or soon thereafter, thereby protecting the egg from desiccation. Pasture habitat, therefore, may be preferred by gallers, as numbers of arthropod predators and parasitoids are lower there than in secondary forest. Parasitoid abundance may also decrease in pasture areas because of the way they chemical stimuli released by the plant its host feeds on (Gauld & Hanson, 1995). Therefore, parasitoids of Mompha sp would be more attracted to secondary growth where more chemical stimuli are released because C. oerstediana is the dominant tree species there. Parasitoids are known to, after attack, have the ability to change the structure of the gall (Hanson, 1995). I f this indeed happened, I would not have been able to record the presence of an attacked gall. With the assumption that parasitoid presence is greater in forest sites, this would contribute to a lower observed abundance of stem galls in secondary growth. For any given host there has been found to be an optimum carrying capacity, which is usually perceived by parasites (the gall physical size (Gauld & Hanson, 1995). The greater size of pasture trees (Fig. 2, 3, 4) m ay indicate a higher carrying capacity and therefore account for the difference in gall abundance between habitats. Another possible explanation for the greater abundance of galls in isolated trees may be the surrounding distribution of conspecifics. Past ure trees are harder to find because they are isolated. However, once they are found, the galling insect will most likely oviposit many of its eggs there because there is nowhere else to lay them. In secondary growth however, there are many C. oerstediana available for oviposition, so the egg laying may be more diffuse. Studies have shown that hosts packed in groups are allocated fewer eggs because they have less exposed surfaces (Gauld & Hanson, 1995). Diffuse oviposition is also known to be preferable bec ause it can help to avoid parasitoid attack. Once a parasitoid locates a host site, it remains in that area for a period of time, searching it for potential hosts (Gauld & Hanson, 1995). Therefore, it would be advantageous to the galling insect to spread o ut its eggs to many host sites, deterring the chance that all of them would be parasitized. Diffuse oviposition is also advantageous to gallers because galls on leaves unoccupied by other galls have been found to have a lower rate of abortion than those th at reside on a leaf with one or more galls (Price, 1984). These results contradict information found in Begon et al. (1990), which states that gall formers increase with the abundance of host plants. One may postulate that gall makers would be able to find a clump of trees occurring together more easily than an individual isolated one, and so the clump would be more susceptible to attack. However, experiments have shown that plants grown at high densities had the same probability of having eggs laid on them as did plants grown at low densities (Rausher & Feeny, 1980). Many of the larvae in the galls sampled from the pasture (77 out of 174) were in their sixth instar (Fig. 12). This result most likely indicates that the galls were created at approximately th e same time. It is generally known that Lepidopterans do indeed have a specific reproducing season. It is also possible that the sixth instar is somehow less
susceptible to mortality than other instars. Developmental stage and size are known to be important factors in host identification for parasitoids (Gauld & Hanson, 1995). These larval stages are reaching a more mature stage that can resist abortion or predator attack In conclusion, this study supports the hypotheses that differences in habitats, such as predator and parasite abundance and host density, would influence stem gall abundance. The tree size hypotheses, however, was not supported by the study, showing tha t DBH, tree height, and number of branches are not directly related to stem gall abundance. Pasture trees are found to be larger, possibly due to reduced competition for resources as compared to secondary growth trees. This study gives insight into gall ma ker behavior as well as plant herbivore interactions. It implies complex interactions between trophic levels, such as the effect on distribution and abundance of a parasitic relationship due to the effect of parasitoids and predators on that parasite. This study also shows that there are more factors than just abiotic conditions that vary between open and forested habitats, and that significantly affect the species within these habitats. This study may have been enhanced if an improved means of viewing the galls was used. Gall presence was hard to observe on trees that were very tall, especially the tops of the trees. For further study, it would be interesting to study the abundance of another type of gall between pasture and secondary forest to determine wh ether different species interactions would give different results. Another interesting study would be to examine other habitats, such as primary and riparian areas, and the abundances of galls in those sites. ACKNOWLEDGEMENTS Thank you to Mauricio Garca who patiently helped me decide on a viable project idea and gave me excellent advice, knowledge, and guidance throughout it. Thanks also to Robert Andrew Rodstrom and Richard F.X. Smith V for helping me to obtain many of the galls used in my study. I am grateful to the Estacin Biolgica de Monteverde for the use of its station and grounds, as well as to Federico Muoz, who kindly allowed me to sample trees from his farm. And finally, a special thanks to my family, mi familia tica, and the CIEE staff and students who supported me in all my efforts.
LITERATURE CITED Aldrich, P.R. & J.L. Hamrick. 1998. Reproductive dominance of pasture trees in a fragmented tropical forest mosaic. Science 281: 103 105. Akimoto, S., Y. Yamaguchi. 1994. Phenotypic selection on the process of gall formation of a Tetraneura aphid (Pemphigidae). Journal of Animal Ecology Vol. 63, No. 3: 727 738. Begon, Harper, Townsend. 1990 Ecology: Individuals, Populations and Communities 2 nd ed. Blackwell Scien tific Publications. Boston. Borror, D.J., C.A. Triplehorn, N.F. Johnson. 1989. An Introduction to the Study of Insects Harcourt Brace College Publishers. Fort Worth, Texas. Evans, H.E. 1984. Insect Biology Addison Wesley Publishing Company, Inc. Readin g, MA. Gauld, I.D., P.E. Hanson. 1995. The parasitoid and its interactions with its host. In: Hanson, P.E., I.D. Gauld. The Hymenoptera of Costa Rica Oxford University Press. Oxford. Graham, S.A. 1995. Gall Makers on Flowers of Cuphea (Lythraceae). Biot ropica Vol. 27, No. 4: 461 467. Haber, W.A., W. Zuchowski, & E. Bello 2000. An Introduction to Cloud Forest Trees Mountain Gem Publications, Monteverde, Costa Rica. P. 81 Hanson, P.E. 1995. Phytophagy in the larval Hymenoptera. In: Hanson, P.E., I.D. Gauld. The Hymenoptera of Costa Rica Oxford University Press. Oxford. Hogue, C.L. 1993 Latin American insects and Entomology University of California Press. Berkeley. Janzen, D.H. 1983. Costa Rican Natural History The University of Chicago Press. Chi cago. Mani, M.S. 1992. Introduction to Cecidology. In J.D. Shorthouse and O. Rohfritsch, eds. Biology of Insect Induced Galls Oxford University Press. New York. Olson, L.M. 1994. Virgin versus disturbed: Which forest allows for greater arthropod biodive rsity? CIEE Summer 1994. Price, P.W. 1984. Insect Ecology John Wiley & Sons, Inc. New York. Price, P.W., G. W. Fernndes, & G.L. Waring. 1987. Adaptive Nature of Insect Galls. Environmental Entomology 16: 15 24. Rausher, M.D., P. Feeny 1980. Herbivory, Plant Diversity, and Plant Reproductive Success: The Effect of Battus Philenor on Aristolochia Reticulata Ecology Vol. 61, No. 4: 905 917. Thompson, J. 1992. Gall Making Insect Strategies in Neotropical Forests. Tropical Ecology and Conservation CIEE