Factors influencing dispersal and population structure of phoretic mites in Centropogon solan i folius ( Campanulaceae) and Columnea spp ( Gesneriaceae) Emily Davis Departments of Biology and Environmental Studies, Whitman College ABSTRACT Phoretic hummi ngbird flower mites of the genus Rhinoseius are nectar thieves of Centropogon solanifolius (Campanulaceae) and Columnea spp (Gesneriaceae) at Monteverde. Dispersal by phoresy is a rare and risky, but nonetheless critical, event in the lives of hummingbir d mites (Colwell and Naeem 1994) This study investigates factors that influence phoretic dispersal by R. colwelli (Mesostigmata : Ascidae) by comparing dispersal from C. solanifolius and Columnea spp Additionally, this study explores the relationships between mite population size and structure in the two pl ant species Flower age, mite population density, sex ratio, an d nectar availability were considered as possible influences on dispersal and populat ion structure. Artificial phoresy experiments wer e performed in the field and collected flowers were analyzed for nectar volume and mite populations Population size was significantly positively correlated to di spersal in both flowers ( p < 0.000001, R 2 = 0.229 for C. solanifolius ; p < 0.000001, R 2 = 0.5 7 for Columnea ) whereas n ectar availability and flower age had no sig nificant effect on dispersal. Population size had a significant, negative effect on proportion of males in a flower for both flowers ( C. solanifolius : p = 0.0041 R 2 = 0.236 ; Columnea : Spearman's Rank p = 0.0002, Rho = 0.9328 ). This may be a result of the increasing tendency for male mites to disperse as population size grows, in order to find flowers with more unmated females. The haystack model of group selection might be implicate d in the mite population structure of these two plant species. RESUMEN Phoretic Â‡caros de colibrÂ’es del genero Rhinoseius son ladrones de nÂŽctar de las especies Centropogon solanifolius ( CampanulÂ‡cea ) y Columnea spp ( GesneriÂ‡cea ) en Monteverde. La d ispersiÂ—n por phoresy es un fenÂ—meno riesgoso y poco comÂœn pero no obstante critico en las vidas de Â‡caros de colibrÂ’es Este estudio investiga los factores que influyen la dispersiÂ—n phoretic de l acaro R. colwell i (Mesostigmata : Ascidae), pa r a compara r la dispersiÂ—n de C. solanifolius y de Columnea spp AdemÂ‡s este estudio explora las relaciones entre el tamaÂ–o de la poblaciÂ—n de Â‡caros y su estructura en las dos especies de planta s La e dad de la flor, la densidad de l a poblaciÂ—n de Â‡caros la s pro porciones de los sexos, y la dispon ibilidad de nÂŽctar fueron consideradas como posibles influencias en la dispersiÂ—n y estructura de l a poblaciÂ—n. Los e xperimentos de pho resy art ificial fueron llevados a cabo en el campo y las flores recogidas fueron anali zadas para determinar el volumen de nÂŽctar y la poblaciÂ—n de Â‡caros El t amaÂ–o de poblaciÂ—n tuvo una correlaciÂ—n significa tiva con la dispersiÂ—n en amba s flores (p < 0.000001, R 2 = 0.229 para C. solanifolius ; p < 0.000001, R 2 = 0. 57 para Columnea ) mientra s la dispon ibilidad de nÂŽctar y la edad de flor no tuvieron un efecto significativo en la dispersiÂ—n. El t amaÂ–o de pobla ciÂ—n tuvo un efecto significativo y negativo en el porcentaje de machos en una flor, para amba s especies de planta s ( C. solanifolius : p = 0.0041, R 2 = 0.236; Columnea : Spearman's Rank p = 0.0002, Rho = 0.9328) Esto p uede ser un resu ltado de la tendencia de los machos a dispersar cuando el tamaÂ–o de la poblaciÂ—n crece para busca r flores con mÂ‡s hembras no fertilizadas El modelo hayst ack de selecciÂ—n de grupo puede ser importante en la estructura de l a poblaciÂ—n de Â‡caros en estas dos especies de plantas.
2 INTRODUCTION Phoresy is a passive dispersal mechanism common among flower inhabiting mites that inhabit spatially and te mporal ly isolated and flowers (Tschapka and Cunningham 2004) The mites must rely on flower visiting animals to colonize their host flowers. For example, in the Neotropics, some 200 species of phoretic mites in the genera Rhinoseius and Proctolaelaps (Mesostigm ata : Ascidae) exploit the mutualism between hummingbirds and the plants they pollinate ( Colwell and Naeem 1994 ). Mites ride within the bird nasal cavity and disembark in response to olfac tory cues. Although mites may move from flower to flower on foot, ( Dobkin 1985 ) movement between inflorescences is almost exclusively by phores y on hummingbirds (Colwell 1983 ). Hummingbird mite life stages include an egg stage, a six legged larva, an eight legged protonymph and then deutonymph, and adult males and females (Co lwell and Naeem 1994 ). All stages feed on nectar, and later stages feed on pollen (Co lwell and Naeem 1994 ) ; the mites mate within the corolla of the host flower. Generation time is about a week which may be shorter than the host longevity, th us necessitating dispersal (Colwell and Naeem 1994). Consequently, flower longevity and phenology greatly affect the development and behavior of individuals, as well as the population dynamics of phoretic mites (Colwell and Naeem 1994 ). Most plants pollinated by hummingbirds in the wet tropics produce only one or two flowers per inflorescence per day ( Colwell and Naeem 1994 ), and most plants pollina ted by traplining hummingbirds produce only one flowering in florescence at a t ime (Colwell and Naeem 199 4 ). Thus, phoresy is a necessity for movement among plants and the general rule for movemen t among infloresce nces (Colwell and Naeem 1994 ). Although phoresy is necessary for dispersal and fundamentally important to nearly all aspects of mite ecology, it is a rare event in the lives of individuals, especially among mite species that occupy long lived flowers (Lara and Ornelas 2002) Mite body size generally increases with flower size (estimated as corolla length), because bigger flowers usually have greater nectar flow and space in side (Colwell and Naeem 1994 ). The more abundant nectar flow and pollen supply of larger flowers also support larger breeding g roup sizes per fl ower (Colwell and Naeem 1994 ). Like other arthopods that live in small, isolated breeding groups, hummingbird flower mites (which are functionally haplodiploid) have female bias ed sex ratios ( Colwell and Naeem 1994). The degree of bias depends upon the species of mite and the typical size of breeding groups (Colwell and Naee m 1994). Mites species that live in smaller breeding groups have more female biased sex ratios, while those in the very biggest flowers have sex ratios that are close to equivalent (Colwell and Naeem 1994). The female biased sex ratio, which is 3:1 in Rh inoseius colwelli (Colwell 1983 ) is thought to be due to either local mate competition or group selec tion under the ha ystack model, both of which relate to spatially st ructured populations and low rates of dispersal. Group selection under the haystack mod el favors an even more biased ratio than does local mate competition ( Colwell 1981) Mite s pecies also vary in the sex ratios of dispersers (Colwell and Naeem 1994 ). For larger breeding group sizes, the proportion of male mites that disperse decr ea ses (Colwell and Naeem 1994 ) because males from smaller breeding groups often have more to gain from dispersa l M ale mites tend to disperse when variation in sex ratio among gro ups makes it worthwhile when a male finds himself in a group with a greater pr oportion of males than usual, it will be beneficial to him to disperse to a group where there may be more females (Colwell and Naeem 1994).
3 Male mites may be able to detect local sex ratios and react behaviorally by seeking groups with more favorable sex ratios (Colwell and Naeem 1994 ) Resource s carcity (nectar and pollen), flower age, and population size may also be factor s inducing mite dispersal. The availability of nectar in flowers is dependent on the nectar production rates over the flower's lifetime. Sanders (2002) found that nectar production in C. solanifolius flowers was highest in younger flowers and declined throughout f lower life and that the mi te population of flowers tracked the increases and decreases in nectar production Munoz e t al. (2004) found that for Rhinoseius colwelli in Bomarea sp inflorescences, mite dispersal during artificial phoresy experiments was significantl y correlated with total population size Gordon (2002) found that for R. colwelli in C. solanifolius in the Monteverde cloud forest, population size and the rate of flower colonization were not significantly influenced by flower age, though there was a trend toward increased proportion of male mites with increased group size, and a trend toward increased propor tion of male mites with increasing flower age. This study also found that flowers were colonized rapidly upon opening and that the population size was highly variable (Gordon 2002). This study examines factors influencing phoretic dispersal in the hummingbird pollinated flowers of Centropogon solanifolius and Columnea spp Additionally, this study examines and compares the population structures and sizes of these two similar flowers. Nectar availability, breeding group size, variability of sex rati os, flower age, and flower phenology/longevity may all play a part in determining population size and structure, what gender disperses, and why. Male mites are expected to be dispersers, since flower size (thus breeding group size) is small in C. solanifo lius and Columnea spp M ore dispersal events are expected in older flowers than in younger flowers. More mites are expected to disperse out of flowers in which most of the nectar has already been consumed or in which nectar is no longer being produc ed due to oncoming senescence. More dispersal events are expected in older flowers because mite population densities are expected to increase with flower age Finally, mite populations are expected to exhibit female biased sex ratios, due to either local mate c ompetition or the haystack model. MATERIALS AND METHODS Study Sites This study was carried out between April 14 t h and May 6 th 2007, in lower montane rain forest (Holdridge 1967) forest of Monteverde, Puntarenas, Costa Rica. Study sites were located in the elfin forest on the ridge in the private reserve of la EstaciÂ—n BiolÂ—gica, on the Sendero s Mirador and Principal 1700 1800 m An additional study site was a treefall in the Bullpen on the Campbell property in Monteverde (1540 m) Study Organis ms Centropogon solanifolius found from 800 2800 m in Costa Rica is hummingbird pollinated and has curved flower tubes clustered in inflorescences at the top of the stem (Zuchowski 2005) This flower is protandrous, passing through a staminate (male, po llen dispersing) stage first, then a pistillate (female, pollen receiving) stage (Zuchowski 2005 ); floral longevity is 7.0 8.4 days (Weiss 1996 ). The plant flowers year round (Kopt ur et al. 1988 in Nadkarni 2000 ).
4 Columnea spp. ( Gesneriaceae) is a gro up of epiphytic, long corolla h ummingbird pollinated flowers (Zuchowski 2005 ). Its flowers, like those of C. solanifolius are long lived and pass through a staminate and pistillate phase (Zuchowski 2005 ). Rhinoseius colwelli (Me sostigmata : Ascidae ) is a phoretic hummingbird mit e (Colwell 1973 ) that inhabits mostly flowers of the genus Centropogon (as well as a species of Columnea ) in the Costa Rican highlands. Its generation time is 7 10 days (Colwell 1973 ). At Monteverde, R. colwelli coexists wi th R. richardsoni a fourth species of Rhinoseius and a species of Proctolaelaps (Colwell 1973 ) Although there was no way of knowing which species of mites were collected from the field, it is most likely that they were exclusively R. colwelli Rhinose ius mites are functionally haplodiploid (Colwell 1973 ) Collection and field experiments A total of 167 flowers were collected, 37 individuals of Columnea spp and 130 of C. solanifolius. Collected flowers were recorded as either accessible to hummingb irds or inaccessible to hummingbirds (with the corolla opening facing the ground or surrounded by dense vegetation), in order to measure whether the mites colonizing the flowers had arrived there by phoresy or on foot. Upon collection, flowers were visual ly examined for mite presence and activity at the lip or on the calyx of the flower An artificial phoresy experiment was performed upon each flower collected with a slender, pointed wooden stick (si mulating a hummingbird beak) inserted into the cor oll a for five seconds Dispersing mites were placed into vials of ethanol for later analysis. Cotton balls with acetone were added to vials with collected flowers to kill mites. Nectar analysis Thirty seven flowers of C. solanifolius were analyzed for nectar volume and concentration. Open flowers free of damage were selected and covered with mesh bags on the morning of the day prior to collection. Samples were analyzed in the laboratory for nectar volume and concentration on the day of collection to pr event changes due to evaporation or reabsorption A 120 microliter micropipette was inserted into each flower and suction applied to remove the nectar. Calipers were used to measure the height of the nectar collected in millimeters, and volume (microlite rs) was calculated using r 2 h To obtain the sucrose concentration (equivalence by weight/total weight of solution) of the nectar, nectar drops were placed on the prism of a percent sucrose Hand Refractometer (Reichert Co.) and percent sucrose by weight was recorded Nectar was not collected in Columnea spp due to time constraints. Laboratory analysis Relative flower ages were measured in both species, since actual flower age could not be determined at the time of collection For C. solanifoli us corolla length was measured and compared to style length, since the style keeps growing and emerging from the anthers during the lifetime of this protandrous flower. In Columnea anthers keep growin g so corolla length was compared to anther length. Ratios were ca lculated to determine the relative age of the flowers. For both species, it was noted whether the flower was in a pistillate or staminate state.
5 Using a dissecting scope, all samples were examined for mite presence inside the flower. F lowers were sliced open and all mites inside were counted, as well as those on the calyx, on the outside of the corolla, or in the vial. Number of mites was counted as well as number of females, males, and nymphs. Nymph s were much smaller, and their sexe s could not be determined Adult males were distinguishable from females by their longer setae and more heavily scleroterized appearance ( Colwell 1983 ) Adult males also tended to be slightly larger and a darker amber color than the white females and nym phs. The sexes of dispersed mites collected in the field were determined with the microscope. Data a nalysis T tests were performed to determine if the average number of total mites and dispersing mites differed between C. solanifolius and Columnea sp p Expected versus observed mi te populations in accessible versus inaccessible flowers were compared between species with a chi squared test. Expected versus observed frequencies of sex classes (male, female, nymph) in the two species were compared with a chi square test Chi square was a lso used to compare expected versus observed frequencies of dispersal events in the two species. Simple regression analyses were used with data from both flowers to exa mine correlations between age and total mite popula tion, age and number o f dispersing mites, dispersal and population size, dispersal and flower age, nectar volume and flower age, nectar volume and dispersing mites, age and number of males, age and number of females, proportion of males and age, and propor tion o f males and population size. One Spearman's Rank test was performed to determine association between proportion of males and population size in Columnea RESULTS Mite population structure A total of 713 mites in 167 flowers were collected (13 0 C. solanifolius and 37 Columnea ). Out of 713 mites counted, very few only 86 dispersed during the artificial phoresy experiments. The mean number of total mites found in any flower was 3.78, with a range of 0 39; the range for C. solanifolius was 0 23 and 0 43 for Columnea (Figure 1)
6 Figure 1. Frequency distribution comparing the number of hummingbird flower mites found in C. solanifolius and Columnea flowers at Monteverde. The population structure is extremely variable. Mites coloniz ed 90 of the 167 flowers collected (54%); mites colonized 20 of 37 collected Columnea (54%) and 70 of 130 C. solanifolius (54%). Though Columnea flowers trended toward a higher number of total mites per flower (including those which dispersed) than C. sol anifolius (Figure 2) the difference in population size was not significant (p= 0 .494, unpaired t test = 0.686, df =165) The mean population size (estimated as total mites within flower plus dispersing mites) for C. solanifolius was 3.6 and 4.7 for Colum nea (standard deviation= 4.9 and 9.7 respectively). Means of dispersing mites between the two species were also not significant (p = 0 .154789, t test =1.429, df =165). Mite population sizes in hummingbird accessible flowers of C. solanifolius were significantly greater than in inaccessible flowers (Figure 2 ) demonstrating the importance of dispersal by phoresy for R. colwelli and the low probability of dispersal on foot (p value = 2.3x10^ 11, Chi square = 44.7, df =1). Figu re 2 Mite populatio ns in accessible versus inaccessible C. solanifolius flowers The populations differed significantly
7 Simple re gressions showed no significant correlation between flower age and total mite population in C. solanifolius (p = 0.56 R 2 = 0 .00274 df = 1.126, beta = 0.052, F = 0.35 ) or Columnea spp (p = 0 .3857, R 2 = 0 .0222 df = 1.34, beta = 0.15, F = 0.77 ). Flower age did not significantly influence numbers of male or female mites in C. solanifolius or the number of male or female mites in Colu mnea spp Nectar volume did not significantly influence total mite population. Factors influencing mite dispersal Out of all flowers collected and 86 dispersing mites, 64 (74.4%) were females, 15 (17.4%) nymphs, and only 7 males (8.14%). (Differences i n sex class dispersal frequency between the two flowers were not tested for significance.) There was no significant correlation between flower age and number of dispersing mites ( C. solanifolius : p = 0 .64 R 2 = .00175, df = 1.126, beta = 0.4, F = 0.22, C olumnea : p = 0.79 R 2 = 0.022, df = 1.35, beta = 0.15, F = 0.79 ). Similarly, nectar volume played no role in number of dispersing mites in C. solanifolius (p = 0.49 R 2 = 0.013, df = 1.36, beta = 0.12, F = 0.48 ). However, mite d ispers al numbers did cor relate significantly with total population size in C. solanifolius (p< .000001, df = 1.125, F= 37.131, R 2 = 0 .229 beta = 0.48 ), and in C olumnea spp (p<.00001, df = 1.35, R 2 = 0.57, F = 46.9, beta = 0.757 ) demonstrating a positive correlation between pop ulation and dispersal (Figure 3 )
8 a) b) Figure 3 a) The relationship between total mite population and number of dispersers in Columnea flowers. Population size was significant ly, positively correlated with dispersa l Note that the x a xis has a different scale than b). b) The relationship between the numbers of dispersers and the total mite population and number of dispersers in C. solanifolius flowers at Monteverde. Population size was significantly, positively correlated with disper sal In a simple regression analysis, n ectar volume correlated significantly with flower age in C. solanifolius (p = 0 .044 df =1.36, R 2 = 0.108, F = 4.36 beta = 0.329 ). Flower age had a negative effect on nectar volume. Chi squared tests showed th e frequency of dispersal events between C. solanifolius and Columnea to be non significant. Mean sex ratio (proportion of mature males) was 0.22 in C. solanifolius and 0.18 in Columnea (standard deviation 0.1713 and 0.118 respectively) The mean sex rat io did not differ significantly between the two flowers (p = 0.95406, df = 1). There was however, a significant difference between the frequencies of each sex class (males, females, and nymphs) found within the total populations of the two species (Figure 4 ), ( p<0. 05, df = 2, chi squared = 27.89).
9 Figure 4 Distribution of total mites found by sex class (male, female, and nymphs, which could not be sexed). Distribution of mite sexes differed significantly between C. solanifolius and Columnea I n simple regression analyses, flower age did not correlate significantly with sex ra tio ( proportion of mature males ) in either Columnea or C. solanifolius For Columnea p = 0.1199, F = 3.1 df = and R 2 = 0.30 ; for C. solanifolius p = 0.66 F = 0.196 d f = R 2 = 0.00628 However, Columnea flowers did show a trend toward significance, with proportion of males decreasing with flower age. Breeding group size, estimated as the total number of mites found within a flower plus its dispersers, correlated signi ficantly to the proportion of mature male mites per flower in both C. solanifolius and Columnea (Figure 5 ). For C. solanifolius a simple regression analysi s showed p = 0.0041, F = 9.58, R 2 = 0.236 For Columnea a Spearman's Rank correlation showed p = 0.0002, Rho = 0.9328.
10 a) b) Figure 5 a) The relationship between mite population of individual flowers and sex ratio (proportion of mature male mites) in C. solanifolius Total mite population has a significant negative correlation with the proportion of ma le mites Flowers that did not contain any mature male mites were excluded from this regression. b) The relationship between mite population of individual flowers and sex ratio (proportion of mature male mites) in C olumnea Total mite p opulation has a significant negative effect on the proportion of mal e mites Flowers that did not contain any mature male mites were ex cluded from this test Although a trendline is included to show correlation, this test was a non parametric Spearman's Rank not a regr ession. Note that the x axis has a different scale than a).
11 Table 1. Summary of significant and non significant factors influencing dispersal, population size, and population structure in C. solanifolius and Columnea spp. Nect ar volume Relative flower age Population size Dispersal No effect No effect Significant, positive correlation in both species Population size No significant effect; negative trend in C. solanifolius No effect ------------Total number of males/female s No effect No effect No effect Proportion males Not tested No significant effect; negative trend in Columnea Significant, negative correlation in both species DISCUSSION Factors influencing mite dispersal As has been demonstr ated in this study (see Figure 2 ), and in the literature, dispersal via phoresy is a rare but necessary event for mite species in long corolla, long lived flowers, and most mites complete their entire life cycle within the flower. Whoever disperses must have benefits to gain fro m doing so Among Rhinoseius mites in C. solanifolius and Columnea spp flowers at Mont everde, i t appears that dis persal is a multifactorial phenomenon, with multiple interacting influences Most dispersing mites were female. This is not what would be expected from mites that live in small groups, in small flowers (Colwell and Naeem 1994 ). In C. solanifolius and Columnea perhaps females have more to gain than males from leaving flowers to found new colonies for reasons that are yet unclear. Howeve r, it is possible that many of the mites categorized as nymphs during the study were in fact subadult males, or that some males were misidentified as females. Nectar volume in C. solanifolius flowers did n ot influence dispersal of mites This is i n agreement with Mun oz et al.'s 2004 finding for mites in Bomarea flowers. Mites can consume up to 40% of a flower's nectar when hummingbi rds are excluded (Colwell 1995), as they were in this study t his may be a confounding factor in the amount of nectar actually gathered from each flower. Another confounding factor might be the "bonanza blank" pattern of nectar production, in which some flowers on an inflorescence will pr oduce no nectar and others a high volume in order to encourage traplining birds to visit more flowers in their foraging efforts (Colwell 1995). This bonanza blank pattern was found in C. solanifolius at Monteverde by Sanders (2002). The small sample size of flowers analyzed for nectar in this study (37 ) may have sampled a biased number o f "blank" flowers. Because the majority of mites complete their life cycle within the
12 flower and nectar availability falls sharply at the end of a C. solanifolius flower's life (Sanders 2002), it seems unlikely that lack of nectar would pr ompt dispersal, especially when considering that a dult mites feed on pollen as well as nectar. Mite dispersal from both species studied was significantly influenced by total mite population within the flower As mite populations rose, so did the likelihood that m ites would disperse, and the total number of dispe rsing mites (see Figure 3 ). This positive correlation upholds Munoz et al.'s 2004 findings. As po pulation sizes rise, chance dictate s that there will simply be more individuals available to disperse Dis persal also carries increased benefits when i ncreased population density also means fewer food resources for mites, although lack of nectar in flowers does not seem to play a role in dispersal. Changing sex ratio as a population grows may also influence d isper sal when populations are high. Relative flower age was not a factor in mite dispersal possibly because flower age is not a good indicator of mite population size or of nectar production Mites were not more likely to disperse in one species of flower versus another Factors influencing population size and structure Mite populations in both plants studied did not differ, but were extremely variable (Figure 1), as in Gordon's 2002 study of population structure in C. solanifolius at Monteverde The variability is probably due to the small size of breeding groups and the infrequency of dispersal, which leads to much stochasticity in population structure (Colwell and Naeem 1994). Although Columnea flowers trended toward higher populations than C solanifolius the difference was not significant The lack of difference ma y be due to similarity in corolla length, which is usually correlated with population s ize (Colwell and Naeem 1994). It is questionable why a trend toward higher mean mite popul ations in Columnea would exist, given this lack of difference in corolla length. It is possible that the variability in population size s might be obscuring a true difference that would be reveal ed through sampling more flowers Relative flower age was not a significant influencing factor on the number of male, female, or total mites in accordance with Gordon's (2002) study A possible reason why no trend s appear in C. solanifolius may be rel ated to the high rates of floral parasitism by a fly larv ae which eats pollen in unopened flowers and affect s the longevity of male and female phases of this plant (Weiss 1996). Parasitism may potentially reduce the accurac y of the relative age estimates. A significant trend might be revealed if actual age wer e measured in days. A large proportion of C. solanifoli us flowers collected were parasitized. There was a strong trend in the correlation of nectar volume with mite population, although this was not significant. Gordon (2002) found that mite populat ions tracked nectar production over a flower's lifetime. Colwell and Naeem (1995) also found mite populations to be positively correlated with nectar production. Here, it appears that increased nectar volume affe cted mite populations negatively but nectar volume and mite populations in this study were extremely va riable. The trend is contradicted by research by Lara and Ornelas (2002), who found that flower mites consume proportionately more nectar in long lived flowers that produce large quantities of ne ctar (such as the flowers analyzed in this study) t han in short lived flowers. This study's finding makes sense when considering that the highest mite populations tended to be found in flowers with low to no nectar; that theoretically, mite populations in the flower will increase with its age; and that the majority of nectar in C. solanifolius is produced in greatest quantity when the flowers are young when mite populations have not increased (Sanders 2002).
13 However, this study and Gordon's (2002) both found that mite populations did not correlate with age, so the find ings may be unaccounted for. Again, a greater sampling effort might have revealed a stronger trend. Factors influencing mite sex ratios Mean proportions of males in both flowers were lo wer than in the published literature both at around 20% mature males, v ersus published accounts of a 3:1 female:male sex ratio for R. colwelli in Colwell (1 973 ). It is possible that the species of Columnea examined ( C. glabra C. magnificans and one unid entifiable other) hosted a different species of mite than R. colwelli and that the sex ratios in Columnea reflect that, but the low sex ratio in C. solanifolius is unaccounted for. Perhaps the species of Centropogon ( C. valerii C. talamancaensis ) previ o usly studied (Colwell 1973 ) are larger flowers that can support larger populations with more males. Although relative flower age was not a significant factor in mite sex ratio (proportion of mature males) for either flower there was a strong trend to ward a smaller proportion of mature males with increasing flower age in Columnea In C. solanifolius there was no trend the inaccuracy of age measuremen ts due to fly parasitism may again be a factor i n confounding a correlation Total population si ze correlated significantly with proportion of mature males in both flower studied, with a strong negative effect of population size on percent males (Figure 5 ) This result contradicts Gordon's 2002 study on Monteverde C. solanifol ius where proportion of males increased with increasing colony size. Additionally, although Colwell and Naeem (1994) state that increased breeding group size should increase the proportion of males, this applies to the average breeding group size of a mite species, not to indiv idual flowers. In other words, the sex ratio bias is not facultative it is a fixed, adaptive response to the average conditions encountered by members of that species (1994) Therefore, there should not be a correlation within one species between group s ize and sex ratio. However, we see one here. There are several possible reasons for this scenario. In this study, population size and sex ratio is extremely variable, with high standard deviation values (nearly equal to the sex ratio values themselv es. This variation in sex ratio would lead to dispersal being especially beneficial for male mites. Males of mites species that live in smaller groups, which have highly female biased sex ratios in flowers (both aspects were demonstrated in this study), are more likely to disperse phoretically than males of species that live in larger groups and have less biased sex ratios (Colwell1994). These small breeding groups are more likely to vary stochastically in their sex ratios. The next inflorescence visite d by a dispersing male might contain more females than males for him to mate with. As time passes and population grows in a flower, a male is likely to mate with the all the females there (age leading to increased females over time, at least in Columnea ), and eventually need to disperse in order to find more unmated females. Thus, more and more males will disperse as time passes, leaving fewer and fewer in the flower, especially since the gravid females therein will produce mostly daughters. Here, we see that continuous dispersal is playing a role in mitigating sex ratios. An alternative reason for the scenario is that if some of the first few colonizers of a flower are males, when the population is small, then the proportion of males will be larger; as time passes and gravid females arrive at the flowers, producing mostly daughters, the sex ratio will become more female biased as it grows. Both alternatives could be functioning together
14 Local mate competition is the model usually invoked to expla in female biased sex ratios in mites. In this model in which the world consists of small "islands," each colonized by a fertilized female, the sex ratio bias is driven by competition between related males over access to mates, often their sisters, leadin g to the evolution of female biased sex ratios to reduce competition among male offspring (Charnov 1982) P arents' fitness is then increased, because competition between male offspring is reduced. This is selection on an individual level: evolution will select for the female who produces the most daughters (Charnov 1982). An alternative and more controversial, model to explain female biased sex ratios is the haystack model, proposed for hummingbird mites by Colwell (1981). In this model, it is gr oup selection that drives the sex ratio bias (Colwell 1981), in randomly structured populations, such as the populations of mites studied here. As in the haystack model, females are the dispersers and colonizers (Charnov 1982). The flower's productivity is related to its population growth; therefore, the more females present, the higher the productivity can be. Population growth is exponential, a trend visible in Figure 3 b). Rather than selecting for individual females, this model selects between group s for those that are most productive (Colwell 1981). The haystack model allows an even more biased sex ratio than local mate competition, and is best applied to small flowers in which the population structure is variable (as seen here) (Charnov 1982). The haystack model is an interesting way to explain this study's results, and may be a better fitting model than local mate competition (LMC). First, it is impossible to know i f LMC is actually functioning in these populations without knowing how many foundresses there were per flower or the actual sex ratio variability of their offspring. The fact that most of the dispersers in artificial phoresy experiments were female could support the LMC hypothesis or the haystack model. However, the greater tha n average female bias found in mite populations of both plant species suggests the haystack model. Further, as seen in Figure 5, the female population appears to build quite a bit as total population grows and proportion of males decreases. Group product ivity increases as the number of females produced increas es, as in the haystack model. Figure 5 also appears to contradict LMC, which predicts an increased number of males with increased group size. Although in both LMC and the haystack model fema les are supposed to be dispersers, in small flowers with variable populations male mites are supposed to have more benefits to gain from dispersal. Dispersing males would essentially "dilute" the forces of selection in both these models. How can these tw o ideas be reconciled? In this study, although Figure. 5 suggests that some males may be dispersing, and thus diluting group selection, the majority of the dispersers were nonetheless female, making up for this effect. However, if local mate competition an d not the haystack model is functioning, Figure 5 suggests a significant dilution of LMC due to dispersing males. Conclusion and future studies Dispersal in R. colwelli populations is a complex phenomenon, influenced by many factors but especially by population size. Overall, there is no difference in dispersal or population structure between C. solanifolius and Columnea Population structure and dispersal often affect each other, and the haystack model may be a good explanation for R. colwelli sex r atios in C. solanifolius and Columnea The observed population size and structure may be impacted by who colonizes flowers first, and the age of flowers when collected and analyz ed. Patterns may be obscured by these impacts. The data from this study sugg est that male individuals of R. colwelli
15 are able to detect variation in the sex ratio of the colonies it inhabits, and disperse accordingly. This has been suggested for other species of mites (Colwell and Naeem 1994). Further studies should focus on improving methodology of estimating flower age and collecting nectar, and increase sample size. A study of the effect on fly parasitism on mite populations in C. solanifolious would be interesting to see if the fly larva's consumption of pollen affects mite populations. Nectar should be studied in Columnea to compare to C. solanifolious A factor not investigated in this study which deserves more attention is whether floral microclimate or other abiotic factors play a role in mite population and dispe rsal in C. solanifolious or Columnea as suggested by Dobkin (1985) Finally, the relationship between population size, proportion of males, proportion of male dispersers, who disperses first, variability in offspring sex, selection, and male ability to d etect sex ratios and react behaviorally should be investigated further to determine whether the haystack model or LMC best explain population structure in R. colwelli at Monteverde. ACKNOWLEDGEMENTS I would like to thank my advisor, Tania Chava r rÂ’a Pizar ro for her continual support, patience, and assistance with statistics. Many thanks also to Drs. Alan and Karen Masters for their inspiration, insights and suggestions throughout the project. To Tom McFarland and Cam ryn Pennington I real ly appreciate you keeping me supplied with snacks, answers to questions, and equipment each day Gracias a mi familia tica, Danis Licho, y Monica, para esta r tan generosos y amable s durante mi permanencia en su hogar Thanks to the EstaciÂ—n BiolÂ—gica for providing me with such an amazing study site. Finally, thank you to my fellow Lower Lab rats for providing laughter commiseration, and conversation during the long hours spent staring at mites under the microscope. LITERATURE CITED Charnov, E.L. 1982. The The ory 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 107(958), 737 760. Colwell, R.K. 1983 Rhinoseius colwelli In: C osta Rican Natural History. D.H. Janzen, ed. 1983. The University of Chicago Press, Chicago, IL, pp. 767 768. Colwell, R.K. 1981. Group selection is implicated in the evolution of female biased sex ratios. Nature 290, 401 404. Colwell, R.K. 1995. Eff ects of nectar consumption by the hummingbird flower mite Proctolaelaps kirmsei on nectar availability in Hamelia patens Biotropica 27(2), 206 217. Colwell R.K. and S. Naeem. 1994. Life history patterns of hummingbird flower mites in relation to hos t phenology and morphol ogy. In: Mites: Ecological and Evolutionary Analyses of Life History Patterns. 1994. M.A. Houck, ed. Chapman & Hall, New York, NY, pp. 23 44. Dobkin, D.S. 1985. Heterogeneity of tropical floral microclimates and the response o f
16 hummingbird flower mites. Ecology 66: 536 543. Fisher, R.A. 1930. Th e Genetic Theory of Natural Selection. Clarendon Press, Oxford. Gordon, S. Population and community structure of hummingbird mites in Centropogon solanifolius (Campa nulaceae). CIEE Spring 2002. Koptur, S., et al 1988. Plants an d vegetation: Seasonality In: Monteverde: Ecology and Conservation of a Tropical Cloud Forest. 2000. N.M. Nadkarni and N.T. Wheelwright, eds. New York University Press, New York, New York p. 66. Holdridge, L.R. 1967. Lifezone ecology Tropical Science Center, S an Jose, Costa Rica. Lara, C. and J.F. Ornelas. 2002. Flower mites and nectar production in six hummingbird pollinated plants with contrasting flower longevities Canadian J ournal of Botany 80, 1216 1229. Munoz, M. C. De Angelo, and A. Paviolo. 2004. MigraciÂ—n de Rhin oseius colwelli en Bomarea sp: U na decisiÂ—n relacionada a l alimento o densidad? La Universidad de Costa Rica y la OrganizaciÂ—n para Estudios Tropicales Eco logia Tropical y ConservaciÂ—n 2004. Sanders, J.B. 2002. Changes in sugar concentration and volume of nectar in Ce ntropogon solanifolius (Campanulaceae). CIEE Spring 2002. Tsc hapka, M. and Saula A. Cunningham. 2004. Flower Mites of Calyptrogyne ghiesbreghtiana (Aracaceae): Evidence for Dispersal Using Pollinating Bats. Biotropica 36: 377 381. Weiss, M.R. 1996. Pollen feeding fly alters floral phenotypic gender in Centropogon solanifolius (Campanulaceae). Biotropica 28(4b), 770 773. Zuchowski, W 2005. A Guide t o Tropical Plants of Costa Rica. Distribuidores Zona Tropical, S.A. Miami, FL, pp. 290, 294 295.
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xlink http:www.w3.org1999xlink xmlns:xsi http:www.w3.org2001XMLSchema-instance
leader 00000nas 2200000Ka 4500
controlfield tag 008 000000c19749999pautr p s 0 0eng d
datafield ind1 8 ind2 024
subfield code a M39-00141
Factores que influyen en la dispersin y la estructura de la poblacin de caros en Centropogon solanifolius (Campanulaceae) y Columnea spp. (Gesneriaceae)
Factors influencing dispersal and population structure of phoretic mites in Centropogon solanifolius (Campanulaceae) and Columnea spp. (Gesneriaceae)
Phoretic hummingbird flower mites of the genus Rhinoseius are nectar thieves of Centropogon solanifolius (Campanulaceae) and Columnea spp. (Gesneriaceae) at Monteverde. Dispersal by phoresy is a rare and risky, but nonetheless critical, event in the lives of hummingbird mites (Colwell and Naeem 1994). This study investigates factors that influence phoretic dispersal by R. colwelli (Mesostigmata: Ascidae), by comparing dispersal from C. solanifolius and Columnea spp. Additionally, this study explores the relationships between mite population size and structure in the two plant species. Flower age, mite population density, sex ratio, and nectar availability were
considered as possible influences on dispersal and population structure. Artificial phoresy experiments were performed in the field and collected flowers were analyzed for nectar volume and mite populations. Population size was significantly, positively correlated to dispersal in both flowers (p < 0.000001, R2 = 0.229 for C. solanifolius; p < 0.000001, R2= 0.57 for Columnea), whereas nectar availability and flower age had no significant effect on dispersal. Population size had a significant, negative effect on proportion of males in a flower for both flowers (C. solanifolius: p = 0.0041, R2 = 0.236; Columnea: Spearmans Rank p = 0.0002, Rho = -0.9328). This may be a result of the increasing tendency for male mites to disperse as population size grows, in order to find flowers with more unmated females. The haystack model of group selection might be implicated in the mite population structure of these two plant species.
Los caros de las flores de los colibres del genero Rhinoseius son ladrones de nctar de las especies Centropogon solanifolius (Campanulcea) y Columnea spp. (Gesnericea) en Monteverde. La dispersin por phoresy es un fenmeno riesgoso y poco comn, pero no obstante critico en las vidas de los caros de los colibres. Este estudio investiga los factores que influyen la dispersin de los acaro R. colwelli (Mesostigmata: Ascidae), para comparar la dispersin de C. solanifolius y de Columnea spp. Adems, este estudio explora las relaciones entre el tamao de la poblacin de caros y su estructura en las dos especies de plantas.
Text in English.
Plant mites--Costa Rica--Puntarenas--Monteverde Zone
Acaros de plantas--Costa Rica--Puntarenas--Zona de Monteverde
Tropical Ecology 2007
Hummingbird flower mites
Ecologa Tropical 2007
Acaros de colibres
Acaros de las flores de colibres
t Monteverde Institute : Tropical Ecology