The Population Structure of Hummingbird Flower Mites in Centropogon solanifolius Campanulaceae Shelley Gordon Department of Exercise Science, University of Puget Sound ________________________________________________________________ ABSTRACT Hummingbird flower mites of the genus Rhinoseius Mesostigmata: Ascidae are nectar thieves of several species of cloud forest plants, including Centropogon solanifolius Campanulaceae. Rhinoseius sp. disperses by riding on the bills of the hummingbirds t hat pollinate the flowers. The population size, flower colonization rate, and sex ratio of Rhinoseius sp. in C. solanifolius were examined in relation to flower age and the distance t o the nearest open neighbor flower NND in Lower Montane Rain Forest in Monteverde, Costa Rica. Population size and the rate of flower colonization were not significantly influenced by the flower age or the NND. Flowers were colonized rapidly and the population size was highly variable. The interaction between flower age and N ND did have a significant effect on the sex ratio of the mite population within a flower Multiple Regression; F = 3.68, p = 0.039. This may be explained by the increased likelihood that female mites leave aging flowers, especially when there are open flo wers nearby. There is also a trend toward a negative relationship between NND and the proportion of mature male mites in a flower, which may be explained by the theory of local mate competition. RESUMEN Ãcaros del genero Rhinoseius Mesostigmata: Ascidae comen nÃ©ctar de algunas especies de plantas en el bosque nuboso, incluso Centropogon solanifolius Campanulaceae. Rhinoseius esp. m ontan en los picos de los colibrÃs para viajar entre las flores. Este proyecto estudiÃ³ el nÃº mero de Ã¡caros, el tiempo cuando llegan a las f lores, y la proporciÃ³n de los Ã¡caros masculinos y femeninos de C. solanifolius en el Bosque Lluviosos Montano Abajo en Monteverde, Costa Rica. Estos facto res fueron examinados en relaciÃ³ n de la edad de las flores y la d istancia entre las flores. El nÃº mero de Ã¡caros y el tiempo cuando llegan a las flores fueron independientes de la edad de las flores y la distancia entre las flores. Los Ã¡caros llegaban a las flores muy rÃ¡pido y el nÃºmero de Ã¡caros en las flores tenÃa mu cha variedad. La interacciÃ³n entre la edad de las flores y la distancia entre las flores tenÃ a un efecto en la proporciÃ³ n de los Ã¡caros masculinos y femeninos RegresiÃ³n MÃºltiple; F = 3.68, p = 0.039. Es posible que este se a porque los Ã¡caros feme ninos vi ajan entre las flores mÃ¡s cuando las flores son mÃ¡s viejas y cerca de otras flores. La teorÃa de competencia entre los masculinos para un cÃ³nyuge puede explicar la relaciÃ³n negativa que se veÃa entre la distancia entre las flores y la proporciÃ³n de Ã¡caros masculinos.
INTRODUCTION The pollination mutualism that exists between hummingbirds and plants in a tropical community is well described Long 1997. However, less is known about other species living in the community that rely on this relationship. Several species of mites, known as hummingbird flower mites, live and reproduce within the corollas of flowers pollinated by hummingbirds Heyneman et al. 1991. These mites, Rhinoseius sp. Mesostigm ata: Ascidae, rob the nectar produced by the flower Colwell 1973, 1995. Although the mites can travel between flowers by foot, they are more often phoretic, dispersing by hitching a ride in the nasal passages of hummingbirds that visit the flowers Colw ell et al. 1974 in Colwell and Naeem 1994. Thus, the mite's exploitation of the mutualism between hummingbirds and the flowers they pollinate is key to their ability to disperse and colonize new flowers. Rhinoseius sp. is commonly found in several species of hummingbird pollinated flowers in the cloud forests of Monteverde, Costa Rica, including Centropogon solanifolius Campanulaceae Colwell 1973; Weiss 1996. Rhinoseius sp. has a generation time of five to seven days, during which time it develops from egg to adult within the flower's corolla Colwell 1973. Centropogon solanifolius is a protandrous herb with a flower life of approximately seven days Stratton 1989. The orange corolla is long and slender, with flaring lobes and a protruding tube formed by the fusion of the filaments and anthers, which enclose the style Weiss 1996. These flowers provide a suitable environment for hummingbird flower mites because they produce nectar that is rich in sucrose, with a Brix measurement of 26.2 Freeman et al . 1985; Sanders, this volume. The flower is also relatively long lived, which may allow the mites to complete the majority of their lifecycle within a single flower Colwell 1995. Centropogon solanifolius tends to grow in patches near trails in the under story of the cloud forest personal observation. The hummingbirds that visit the flowers are trapliners with a consistent flower visitation pattern Colwell 1973; Stiles and Skutch 1989; Long 1997. A flower that is surrounded closely by other flowers wou ld theoretically be more attractive to hummingbirds because multiple flowers translate to increased nectar availability. The number of hummingbird visits to a flower patch influences the likelihood of colonization by mites. Thus, flower colonization by mit es is likely to be affected by hummingbird behavior as mediated by flower patch size. Hummingbird flower mites are functionally haplodiploid haploid males and diploid females and have an adult sex ratio of two to three females per male Colwell 1983; Nor ton et al. 1993. However, as the population size increases, the female biased sex ratio decreases Wilson and Colwell 1981; Colwell and Naeem 1994. This phenomenon is explained by the local mate competition LMC model, which states that a female increases her fitness by reducing the number of male offspring when colonies are small to prevent mate competition between her sons Charnov 1982. As the colony size increases, LMC is reduced and the sex ratio moves toward 50:50, as predicted by Fisher's model Fisher 1930 in Wilson and Colwell 1981. However, a group selection model predicts that even in large colonies there is selection for a female biased sex ratio because within the entire population there is competition for colonization sites Wilson and Colwell 1981. Since females comprise the dispersal stage in hummingbird flower mites, a given female will be more likely to increase her fitness by producing more daughters and monopolizing colonization sites Kaliszewski and Wrensch 1993. T herefore, the population size of hummingbird flower mites can have an effect on the adult sex ratio.
This study investigated the population size, sex ratio, and colonization rate of newly opened C. solanifolius flowers by hummingbird flower mites. The foll owing questions about mite populations were addressed: 1 How does nearest neighbor distance NND and flower age influence the number of mites in an individual flower? 2 What are the effects of NND and flower age on the proportion of adult male mites i n a flower? 3 What is the effect of flower age on the frequency of colonization? As the flower aged, it was predicted that the population of mites would increase and the sex ratio would become less female biased because time should allow for colonization , as well as the production of offspring. In addition, a small NND should have a larger population size and a less female biased sex ratio than a flower with a large NND. Likewise, the LMC and group selection models predict that the proportion of adult mal e mites should be negatively correlated with NND, however the sex ratio should never reach 50:50. The proportion of flowers colonized was expected increase as the flower aged because time should allow for increased discovery by mites. MATERIALS AND METHODS This study was conducted in the private cloud forest reserve above the EstaciÃ³n BiolÃ³ gica Monteverde, Puntarenas, Costa Rica. Data were collected from April 12 to May 7, 2002. The study sites were located next to trails through the Lower Montane Ra in Forest, at an elevation of 1700 1800 m. Ninety six unopened C. solanifolius flowers were identified, labeled, and covered with a mesh net designed to prevent hummingbird visits. Other open flowers on the same inflorescence were removed prior to covering the selected flower in order to prevent the likelihood of mite transfer between flowers prior to mesh net removal. However, flowers on different inflorescences were not removed or altered so that the attraction of the hummingbird to the plant was not dist urbed. When the flower opened, the mesh net was removed and the flower was randomly assigned a collection date. The date of collection corresponded to a particular flower age, which was determined by the number of days the flower had been open day one d ay seven. On the assigned day the flower was collected and placed into a small plastic vial. Two drops of acetone were added to the vial at the time of flower collection in order to prevent mite escape. The distance to the closest open flower NND was me asured in meters and recorded for each collected flower. Each flower was viewed under a dissecting microscope. The number of mites and the proportion of mature male mites it contained were recorded. Mature males were more sclerotized, which allowed them to be identified under a microscope because they appeared darker in color. The stage of flower development either staminate or pistillate was also recorded. Twenty flowers were collected on days one, two, and three, and 21 were collected on day four. Due t o time constraints, only six flowers were collected on day five, four on day six, and five on day seven. Data Analysis Flower age was estimated as the number of days it was open upon collecting. The number of mites per flower was sine transformed and the NND values were log transformed to normalize the data. By excluding all flowers that lacked males and days five, six and seven, the data for the proportion of mature male mites was normalized. A multiple regression analysis was conducted to determine if there was an effect of flower age and the NND on the
mite population size or the proportion of mature males. In addition, an unpaired t test was conducted to determine the relationship between the stage of flower development male vs. female stage and the number of mites in the flower. The number of mites per flower was compared to the proportion of mature males with a simpl e regression analysis. A seven by two contingency table was conducted using a chi squared test to examine the effect of flower age on the frequency of flower colonization by mites. RESULTS The mean number of mites found in a flower was 7.18 Â± 8.27, with a range of 0 37 Figure 1. Mites colonized 69 of the 96 sampled flowers 72%. There was no significant effect of flower age or NND on the mite population size within a flower Multiple Regression; N = 96, R 2 = 0.019, F = 0.917, p = 0.403; see also Figure 3. The relationship between the number of mites present in the flower and flower sex was not significant male flowers, X M = 0.035 Â± 0.374, N M = 53; female flowers, X F = 0.121 Â± 0.387, N F = 42; Unpaired t test; t = 0.674, p = 0.502. When the population of mites in a flower included mature males, a mean of 2.00 Â± 1.10 males and a mean proportion of 0.18 Â± 0.11 males were found Figure 2. There was not a strong relationship between the number of mites per flower and the flower's propor tion of mature males Simple Regression; N = 29, R 2 = 0.048, F = 1.34, p = 0.257; Figure 4. The interaction between flower age and NND had a significant effect on the proportion of mature males Multiple Regression; N = 29, R 2 = 0.221, F = 3.68, p = 0.039. However, when flower age and NND are considered alone for their effect on the proportion of male mites, the effects are not significant t = 1.73, p = 0.095; t = 1.66, p = 0.109; respectively; Figures 5 and 6. There was not a significant trend se en between the probability of flower colonization by mites and the flower age Chi squared test; X 2 = 7.10, df = 6, p = 0.311; Table 1; Figure 7. DISCUSSION Population Size The high variability in the data for mite population size made it difficult to as sess the effects of flower age and NND on hummingbird flower mite populations. Although a multiple regression revealed that the interaction between flower age and NND did not have a significant effect on the population size, the relationship between the n umbers of mites and NND reveals and interesting trend Figure 3. Mites almost always colonized flowers that had a close neighbor. The dashed oval highlights the absence of flowers that had a small NND and lacked mites. Two factors may be responsible for t his finding. First, mites do not rely solely on hummingbirds for dispersal. They also move on foot between successive flowers Colwell 1983. Second, a small NND would suggest that the hummingbird attraction would be enhanced as a result of increased nectar availability in a small area. These factors indicate that a flower with a small NND should have a higher probability of being rapidly and repeatedly colonized, thus explaining the lack of flowers that had close neighbors and no mites. As a flower ages, it would be expected that the number of mites would also increase because time allows for reproduction and repeated colonization to occur. In this study there
was no significant rela tionship found between flower age and population size. These results are consistent with those of a previous study conducted by Colwell 1973. He compared the mite population size to the stage of flower development either staminate or pistillate and fou nd that the number of mites was not dependent on flower age. The data Colwell collected were also highly variable in that mite populations were large in some flowers and relatively small in others. The most likely reason for this finding was that some fact or prevented the mite population from reaching its carrying capacity in some flowers. He found that predation by a hemipteran was common in hummingbird flower mite populations and suggested that this could effectively prevent the population from reaching t he carrying capacity of the flower. Mite predation was not measured in this study, however in several cases there were unidentified insect species present in collected flowers. Further research could be done to determine if mite population growth is someho w limited before the carrying capacity of the flower is reached. Sex Ratio In most populations, a 50:50 sex ratio is selected for because it maximizes the reproductive success of both the males and females Fisher 1930. However, species that exhibit haplodiploidly and local mating often produce a female biased sex ratio. In a hummingbird mite population, local mating occurs between siblings prior to dispersal Wilson and Colwell 1981. In a colony formed by one foundress, the maximum fitness o f the female is achieved when the male offspring number is reduced, and the female's energy is re allocated to the production of additional female offspring. This reduces local mate competition among male siblings and maximizes the number of possible dispe rsers. However, as the number of foundresses in a local community increases, the selective pressure to produce a female biased sex ratio decreases accordingly Charnov 1982. The local mate competition LMC model suggests that as the size of a local commu nity increases, the proportion of males will increase. The model of group selection defines the sex ratio of a population as a dynamic equilibrium between differing local and community selective pressures Wilson and Colwell 1981. It considers that althou gh an increase in local population size will select for a less biased sex ratio, the population wide competition for colonization sites selects for the colony to produce as many dispersers as possible. Since mating in a mite population occurs prior to disp ersal, mated females are the primary dispersers Kaliszewski and Wrensch 1993. Thus, a female biased sex ratio is selected for on the community level. This study did not show a strong relationship between colony size and the proportion of male mites as wo uld be predicted by LMC and group selection Figure 4. However, there was a trend toward a negative relationship between NND and the proportion of mature male mites in the colony Figure 6. As the distance between open flowers decreased, the proportion o f males increased, but never reached 0.5. This supports both the LMC and group selection models. There was also a positive trend in the relationship between flower age and the proportion of mature males Figure 5. Male mites are infrequent dispersers; the refore increased dispersal by males is unlikely to be responsible for this trend Colwell and Naeem 1994. Mites in immature stages cannot be sexed, and development from egg to adult form requires only two to three days Colwell 1973; Wilson and Colwell 19 81. As the flower aged, it is possible that the proportion of males showed an increase because time allowed more male mites to reach maturity.
The interaction between flower age and NND did have a significant effect on the proportion of mature male mites living within a flower. Female dispersal could possibly justify these findings. Adult females comprise the dispersal stage of a mite population Kaliszewski and Wrensch 1993. Both a small NND and an older flower would increase the likelihood that a mature female would leave the flower. If this were the case, an increase in the proportion of male mites in older flowers with close neighbors would b e expected. Colonization Frequency Flower colonization is dependent on the rate of discovery. The data from this study did not reveal a significant difference in the probability of colonization as the flower aged. However, there was a high flower discovery rate by the mites Figure 7. On day one 60% of the flowers were colonized. As the flower aged there was a trend toward an increased proportion of colonized flowers, and by day six all of the flowers had been colonized. These results can be explained by t he blooming pattern of C. solanifolius. Centropogon solanifolius exhibits asynchronous flowering throughout the year, providing a constant habitat for Rhinoseius sp. Colwell 1973; Weiss 1996. When a flower senesces, the mites are forced to disperse to ne w ones either by walking or by riding on a hummingbird. Since blooming of C. solanifolius is continuous, there is a constant flow of mites dispersing and colonization of newly opened flowers is rapid Wilson and Colwell 1981. Conclusions The results of this study suggest that the population size of hummingbird flower mites in C. solanifolius is highly variable, and thus is not significantly affected by NND and flower age. The proportion of mature male mites was influenced by the interactio n between NND and flower age, which suggests mites in an aging flower that has a small NND are more likely to disperse. The colonization of a flower by mites was rapid. This may be explained by the asynchronous flowering exhibited by C. solanifolius, which caused there to be a constant flow of mites into the dispersal pool. ACKNOWLEDGMENTS I would like to thank Karen Masters for her help in project design and data analysis, and also for brightening every day with her love of life. Many thanks to Will Wieder and Andrew Rodstrom for their countless hours of assistance with everything and consistent encouragement. I am also grateful to the family of Maria Edith Abarca and Norman Santa Maria for their support throughout the duration of this project. An additional thank you to Jeff Sanders for also working on C. solanifolius up in the elfin forest and for helping pull it all together. I want to thank everyone at CIEE this semester for all their hard work, but also for always finding time to enjoy it as well. _____________________________________________________________________________________________
L ITERATURE CITED Charnov, E.L. 1982. The Theory of Sex Allocation. Princeton University Press: Princeton, NJ, pp. 67 92. Colwell, R.K. 1973. Competition and coexistence in a simple tropical community. American Naturalist. 107958, 737 760. Colwell, R.K. 1983. Rhinoseius colwelli. In: Costa Rican Natural History. D.H. Janzen, ed. The University of Chicago Press, Chicago, IL, pp. 767 768. Colwell, R.K. 1995. Effects of nectar consumption by the hummingbird flower mite Proctolaelaps kirmsei on nectar availability in Hamelia patens. Biotropica. 272. 206 217. Colwell, R.K., B.J. Betts, P. Bunnell, F.L. Carpenter, and P. Feinsinge r. 1974. Competition for the nectar of Centropogon valerii by the hummingbird Colibri thalassinus and the flower piercer Diglossa plumbea. And its evolutionary implications. Condor. 76. 447 452. Colwell, R.K. and S. Naeem. 1994. Life history patterns of hummingbird flower mites in relation to host phenology and morphology. In: Mites: Ecological and Evolutionary Analyses of Life History Patterns. M.A. Houck, ed. Chapman & Hall, New York, NY, pp. 23 44. Fisher, R.A. 1930. The Genetical Theory of Natural Selection. Clarendon Press, Oxford. Freeman, C.E., R.D. Worthington, and R.D. Corral. 1985. Some floral nectar sugar compositions from Durango and Sinaloa, Mexico. Biotropica. 174. 309 313. Heyneman, A.J., R.K. Colwell, S. Naeem, D.S. Dobkin, and B. Hallet. 1991. Host plant discrimination: Experiments with hummingbird flower mites. In: Plant Animal Interactions: Evolutionary Ecology in Tropical and Temperate Regions. P.W. Price, T.M. Lewinsohn, G.W. F ernandes, and W.W. Benson, eds. John Wiley & Sons, Inc., New York, NY, pp. 455 485. Kaliszewski, M. and D.L. Wrensch. 1993. Evolution of sex determination and sex ratio within the mite cohort Tarsonemina Arcari: Heterostigmata. In: Evolution and Diversity of Sex Ratio in Insects and Mites. D.L. Wrensch and M.A. Ebbert, eds. Chapman & Hall, New York, NY, p. 192 213. Long, K. 1997. Hummingbirds. A Wildlife Handbook. Johnson Books: Boulder, CO. Norton, R.A., J.B. Kethley, D.E. Johnston, and B.M. O'Connor. 1993. Phylogenetic perspectives on genetic systems and reproductive modes of mites. In: Evolution and Diversity of Sex Ratio in Insects and Mites. D.L. Wrensch and M.A. Ebbert, eds. Chapman & Hall, New York, NY, p. 17. Sanders, J.B. 2002. Changes in sugar concentration and volume of nectar in Centropogon solanifolius Campanulaceae. CIEE Spring 2002. Stiles, F., and A. Skutch. 1989. A Guide to the Birds of Costa Rica. Cornell University Press: Ithaca , New York. Stratton, D.A. 1989. Longevity of individual flowers in a Costa Rican cloud forest: ecological correlates and phylogenetic constraints. Biotropica. 214. 308 318. Weiss, M.R. 1996. Pollen feeding fly alters floral phenotypic gender in Centropogon solanifolius Campanulaceae. Biotrop ica. 284b. 770 773. Wilson, D.S., and R.K. Colwell. 1981. Evolution of sex ratio in structured demes. Evolution. 355. 882 897.
Table 1. Chi squared test results for the effect of flower age on the probability of flower colonization by mites. X 2 = 7.098, df = 6, p = 0.3119. Day Y N Tot al 1 12 8 20 2 14 6 20 3 16 4 20 4 17 4 21 5 4 2 6 6 4 0 4 7 2 3 5 Tota l 69 27 96 Figure 1. Frequency distribution of the number of hummingbird flower mites found in Centropogon solanifolius flowers. N flowers =96, x mites = 7.18 Â± 8.27.
Figure 2. Frequency distribution of the proportion of mature male hummingbird flower mites found in Centr opogon solanifolius flowers. N flowers = 29, x prop males = 0.18 Â± 0.11. Figure 3. The relationship between the nearest neighbor distance and the number of mites found in the flower. The dashed oval highlights the absence of flowers that have a close neighbor and a lack of mites, suggesting that flowers with a small NND are generally colonized by mites. N = 96.
Figure 4. A simple regression graph of the number of mites in a flower and the proportion of mature males. N = 29, RÂ² = 0.048, F = 1.344, p = 0.256. Figure 5. The relationship between the proportion of mature male mites found in a flower and the flower age. Flowers that contained no mature male mites were excluded from this regression, N = 29.
Figure 6. The relationship between the proportion of mature male mites found in the flower and the nearest neighbor distance. Flowers that did not contain any mature male mites were excluded from this regression. Figure 7. Proportion of flowers that are colonized in relation to flower age. N 1 =20, N 2 =20, N 3 =20, N 4 = 21 N 5 =6 N 6 = 4 N 7 = 5.