Transmission of a pathogen in Bombus terrestris,with a note on division of labour in social insects

Summary Parasites of social insect workers can be transmitted within the colony to other, related host individuals or, alternatively, to unrelated workers of other colonies. Division of labour affects the probability of transmission, as young individuals often work inside the nest whereas older ones often leave the nest to forage. Therefore, the relative probabilities of transmission within-vs. between-nests is also affected by the delay between host infection and the shedding of propagules, i.e. the latent period of the parasite strain. We therefore hypothesized that strains of the flagellate parasite Crithidia bombi (Trypanosomatidae, Zoomastigophorea) infecting workers of the bumble bee Bombus terrestris (Hymenoptera, Apidae) could differ in their delays and coexist in a population. This would be the case if strains that are shed after a short time delay were more efficiently transmitted to other colony members, whereas strains with long delays were more efficiently transmitted to non-related workers in the population. We tested this hypothesis by experimentally varying time delay and by allowing transmission to either sister workers from the same nest or unrelated workers from other nests. Transmission of C. bombi was measured as the number of parasitic cells shed by the exposed workers after a standard period. The results showed that relatedness as such had no effect, but that delay and nest identity were highly significant effects to explain variation in transmission success. There was a significant interaction between nest identity and delay, such that bees of some colonies acted as efficient transmitters for C. bombi under short delays and vice versa. We discuss how division of labour may affect parasitism in social insects and, vice versa, how division of labour may be under selection from the effects of parasitism, using available evidence from the literature. Correspondence to: P. Schmid-Hempel     

Effects of Invasive Parasites on Bumble Bee Declines

Abstract: Bumble bees are a group of pollinators that are both ecologically and economically important and declining worldwide. Numerous mechanisms could be behind this decline, and the spread of parasites from commercial colonies into wild populations has been implicated recently in North America. Commercial breeding may lead to declines because commercial colonies may have high parasite loads, which can lead to colonization of native bumble bee populations; commercial rearing may allow higher parasite virulence to evolve; and global movement of commercial colonies may disrupt spatial patterns in local adaptation between hosts and parasites. We assessed parasite virulence, transmission mode, and infectivity. Microparasites and so‐called honey bee viruses may pose the greatest threat to native bumble bee populations because certain risk factors are present; for example, the probability of horizontal transmission of the trypanosome parasite Crithidia bombi is high. The microsporidian parasite Nosema bombi may play a role in declines of bumble bees in the United States. Preliminary indications that C. bombi and the neogregarine Apicystis bombi may not be native in parts of South America. We suggest that the development of molecular screening protocols, thorough sanitation efforts, and cooperation among nongovernmental organizations, governments, and commercial breeders might immediately mitigate these threats.     

Effects of parasitic mites and protozoa on the flower constancy and foraging rate of bumble bees

Parasites can affect host behavior in subtle but ecologically important ways. In the laboratory, we conducted experiments to determine whether parasitic infection by the intestinal protozoan Crithidia bombi or the tracheal mite Locustacarus buchneri alters the foraging behavior of the bumble bee Bombus impatiens. Using an array of equally rewarding yellow and blue artificial flowers, we measured the foraging rate (flowers visited per minute, flower handling time, and flight time between flowers) and flower constancy (tendency to sequentially visit flowers of the same type) of bees with varying intensities of infection. Bumble bee workers infected with tracheal mites foraged as rapidly as uninfected workers, but were considerably more constant to a single flower type (yellow or blue). In contrast, workers infected with intestinal protozoa showed similar levels of flower constancy, but visited 12% fewer flowers per minute on average than uninfected bees. By altering the foraging behavior of bees, such parasites may influence interactions between plants and pollinators, as well as the reproductive output of bumble bee colonies. Our study is the first to investigate the effects of parasitic protozoa and tracheal mites on the foraging behavior of bumble bees, and provides the first report of Crithidia bombi in commercial bumble bees in North America.     

Female great tits do not alter their yolk androgen deposition when infested with a low-transmittable ectoparasite

In birds as in many other taxa, parasites can have deleterious effects on offspring development. Therefore, avian mothers have evolved responses to counteract parasite virulence in offspring via transgenerational defense mechanisms that is the transfer of immune-enhancing substances such as antibodies to their eggs. Another maternal pathway is suggested by the finding that infested great tit mothers produced eggs with lower androgens, since these yolk androgens are immunosuppressive and potentially affect parasite susceptibility of the nestlings. However, whether this pathway is a specific adaptation to infestation with parasites that affect the offspring or an epiphenomenon of lower androgen production in the female due to the parasite effects on the mother itself is as yet unclear. In this study we infested female great tits (Parus major) with sheep ticks (Ixodes ricinus), which are nonnidicolous ectoparasites with low vertical transmission capability, and evaluated the effects on yolk androgen deposition. Tick-infested females did not significantly reduce their deposition of androgens (androstenedione (A₄) and testosterone) compared to tick-reduced females, which is in contrast to a previous study showing a lowered deposition of A₄ and testosterone when females were exposed to the nidicolous hen flea. Thus, females alter their hormone deposition, and thus likely offspring phenotype, when exposed to parasites that also form the parasitic environment of their offspring, but not when temporarily infested with the field-dwelling sheep ticks with low transmission capability. This suggests that selection favored the evolution of an adaptive transgenerational effect by acting mainly on the parasite-induced maternal effect.     

Colony fission affects kinship in a social insect

Establishment of new groups is an important step in the life history of a social species. Fissioning is a common mode not only in group proliferation, for instance, as a regular part of the life cycle in the honey bee, but also when multiple females reproduce in the same group, as in multiple-queen ant societies. We studied the genetic consequences of fissioning in the ant Proformica longiseta, based on DNA microsatellites. In P. longiseta, new nests arise by fissioning from the old ones when they grow large, and the daughter nests consist of workers and queens or queen pupae but never both. Our results show that fissioning is not entirely random with respect to kinship. Workers tend to segregate along kin lines, but only when the initial relatedness in the parental nests is low. Workers in a daughter nest also tend to be associated with closely related adult queens, whereas such an association is not detected between workers and queen pupae. Most queens and workers are carried to the daughter nest by a specialized group of transporting workers, suggesting active kin discrimination by them. Fissioning pattern in P. longiseta is different from that found in other social insects with regular fission (e.g., the honey bee, swarm-founding wasps), where no fissioning along kin lines has been found. It does, however, resemble fissioning in another group of social animals: primates.     

Some notes on the population dynamics of the monogenean gill parasite Gastrocotyle indica

The fish Caranx kalla of different size groups (2 to 18 cm body length) is present during various calender months of the year off the Madras coast (S. India). C. kalla displays a definite progressive shift in growth during successive months; the fish grows from 2 cm (September) to about 18 cm (August) over a period of 12 months. The highest (47%) or lowest (11%) incidence of a monogenean gill parasite Gastrocotyle indica has been observed in the C. kalla size groups 2.0 to 3.9 cm or 16.0 to 17.9 cm; however, the maximum specific density of 10.9 parasite/infested fish was observed in the size group 14.0 to 15.9 cm. Small fish (2.0 to 3.9 cm) carry only larvae (58/100 fish); larger hosts (4.0 to 5.9 cm) larvae (45/100 fish) and juveniles (8/100 fish); individuals of the 6.0 to 15.9 cm size group harbour larval, juvenile and adult stages of G. indica. Fishes of more than 16 cm carry only adult parasites. The level of parasitization is higher on the male hosts (30% incidence; 5 parasites/infested male) than on the females (25% incidence; 4 parasites/infested female). The highest incidences of larval, juvenile and adult G. indica occur in September, February, and May, respectively; the parasite seems to live for 1 year. The host requires a period of about 8 months to acquire immunity against infestation by G. indica or copepod parasites.     

Evidence for differential selection and potential adaptive evolution in the worker caste of an inquiline social parasite

Social parasites exploit the socially managed resources of social insect colonies in order to maximise their own fitness. The inquilines are among the most specialised social parasites, because they are dependent on being fully integrated into their host's colony throughout their lives. They are usually relatives of their host and so share ancestral characteristics (Emery's rule). Closely related inquiline-host combinations offer a rare opportunity to study trade-offs in natural selection. This is because ancestral adaptations to a free-living state (e.g. the production of a worker caste) become redundant and may be replaced by novel, parasitic traits as the inquiline becomes more specialised. The dynamics of such processes are, however, unknown as virtually all extant inquiline social parasites have completely lost their worker caste. An exception is Acromyrmex insinuator, an incipient permanent social parasite of the leaf-cutting ant Acromyrmex echinatior. In the present study, we document the size distribution of parasite and host workers and infer how selection has acted on A. insinuator to reduce, but not eliminate, its investment in a worker caste. We show that the antibiotic producing metapleural glands of these parasite workers are significantly smaller than in their host counterparts and we deduce that the metapleural gland size in the host represents the ancestral state. We further show experimentally that social parasite workers are more vulnerable to the general insect pathogenic fungus Metarhizium than are host workers. Our findings suggest that costly disease resistance mechanisms are likely to have been lost early in inquiline evolution, possibly because active selection for maintaining these traits became less when parasite workers had evolved the ability to exploit the collective immune system of their host societies.     

Effects of kinship on territorial conflicts among groups of lions,<Emphasis Type="Italic"> Panthera leo</Emphasis>

Inclusive fitness theory predicts that cost of tolerant behaviour during competitive interactions is lower for relatives than for nonrelatives. Many studies have examined the effect of relatedness on behaviour within social groups. In contrast, kin selection acting among groups has received less attention. The genetic structure of African lion (Panthera leo) populations creates a strong possibility that kin selection among groups modifies behaviour during group conflicts. We used playback experiments and genetic data to investigate the importance of relatedness during simulated territorial disputes in lions. However, we found no effect of relatedness on territorial behaviour. Degree of relatedness did not affect the decision to approach simulated intrusions, nor did it affect the behaviour during approaches. The decision to approach was instead affected by position within the territory and consecutive playback number (a measure of habituation). For playbacks that did elicit an approach, the speed of response was not detectably affected by relatedness, but was affected by odds (the ratio of residents to intruders), number of intruders, number of bouts, presence of cubs, position within the territory, temperature and playback number. Although responses were unaffected by relatedness, it remains possible that other aspects of behaviour during natural encounters among prides are affected by kin selection.Communicated by L. Sterck     

Kin discriminators in the eusocial sweat bee <Emphasis Type="Italic">Lasioglossum malachurum</Emphasis>: the reliability of cuticular and Dufour’s gland odours

The ability to discriminate degrees of relatedness may be expected to evolve if it allows unreciprocated altruism to be preferentially directed towards kin (Hamilton in J Theor Biol 7:1–16, 1964). We explored the possibility of kin recognition in the primitively eusocial halictid bee Lasioglossum malachurum by investigating the reliability of worker odour cues that can be perceived by workers to act as indicators of either nest membership or kinship. Cuticular and Dufour’s gland compounds varied significantly among colonies of L. malachurum, providing the potential for nestmate discrimination. A significant, though weak, negative correlation between chemical distance and genetic relatedness (r = −0.055, p < 0.001) suggests a genetic component to variation in cuticular bouquet, but odour cues were not informative enough to discriminate between different degrees of relatedness within nests. This pattern of variation was similar for Dufour’s gland bouquets. The presence of unrelated individuals within nests that are not chemically different from their nestmates suggests that the discrimination system of L. malachurum is prone to acceptance errors. Compounds produced by colony members are likely combined to generate a gestalt colony chemical signature such that all nestmates have a similar smell. The correlation between odour cues and nest membership was greater for perceived compounds than for non-perceived compounds, suggesting that variability in perceived compounds is a result of positive selection for nestmate recognition despite potentially stabilising selection to reduce variability in odour differences and thereby to reduce costs derived from excessive intracolony nepotistic behaviour.     

How the social parasitic bumblebee <Emphasis Type="Italic">Bombus bohemicus</Emphasis> sneaks into power of reproduction

Social parasitism is widespread in many groups of social living hymenopteran species and has also evolved in the genus Bombus. Cuckoo bumblebees (subgenus Psithyrus) are obligate brood parasites in nests of other bumblebee species. After nest usurpation and the killing of the host queen, the parasite female has to control worker reproduction in order to accomplish and maintain reproductive dominance and to ensure her reproductive success. The aim of our study was to examine whether the generalist parasitic bumblebee Bombus bohemicus monopolizes and prevents worker reproduction by physical or chemical means and to identify possible odor compounds involved therein. We performed bioassays with callow workers of the host Bombus terrestris and have shown that B. bohemicus females are able to suppress host worker ovarian development, when these host workers are under the direct influence of the parasite female. Furthermore, by chemical analyses, we have demonstrated that the parasite females adjust to the odor profiles of their host queens in order to maintain the level of fertility signaling inside the host colony although the host queen is absent. We also found that host workers change their odor profile after nest usurpation by the parasite female and consequently, we suggest that the host and parasite are caught up in a chemical arms race.     

Weak specialization of workers inside a bumble bee (<Emphasis Type="Italic">Bombus impatiens</Emphasis>) nest

Division of labor is common across social groups. In social insects, many studies focus on the differentiation of in-nest and foraging workers and/or the division of foraging tasks. Few studies have specifically examined how workers divide in-nest tasks. In the bumble bee, Bombus impatiens, we have shown previously that smaller workers are more likely to feed larvae and incubate brood, whereas larger workers are more likely to fan or guard the nest. Here, we show that in spite of this, B. impatiens workers generally perform multiple tasks throughout their life. The size of this task repertoire size does not depend on body size, nor does it change with age. Further, individuals were more likely to perform the task they had been performing on the previous day than any other task, a pattern most pronounced among individuals who guarded the nest. On the other hand, there was no predictable sequence of task switching. Because workers tend to remain in the same region of the nest over time, in-nest workers may concentrate on a particular task, or subset of tasks, inside that region. This division of space, then, may be an important mechanism that leads to this weak specialization among in-nest bumble bee workers.     

Generalist cuckoo bees (Hymenoptera: Apoidea: Sphecodes) are species-specialist at the individual level

Intensive and incessant arms races between a parasite and its host are generally expected to lead to parasite specialization. Nevertheless, some parasitic species still successfully attack wide spectra of hosts. One of the solutions to the evolutionary enigma of the long-term existence of generalist parasites is their specialization at an individual level, a phenomenon well known, e.g., in European common cuckoo. Over its range, it parasitizes a number of bird species; however, individual females are mostly specialists possessing adaptations to a particular host species. In this study, we test the possibility of individual specialization in generalist cuckoo bees, the insect counterparts of avian cuckoos. Females of cuckoo bees lay each egg into a single brood cell in the nests of other bee species. The host’s offspring is destroyed by the parasitic female or later by her larvae, which feed on pollen supplies accumulated by the host. Both studied cleptoparasitic bees (Sphecodes ephippius and Sphecodes monilicornis) are widely distributed in Europe, where they have been reported to use broad host spectra. We recorded several host species (including some previously unknown) for both cuckoo bee species, and confirmed that these parasites are indeed generalist even at a small local scale. However, we demonstrate that exactly as in the avian cuckoos, each female in both species of generalist bee parasites tends to attack just one host species.     

Lack of detectable nepotism in multiple-queen colonies of the fire ant Solenopsis invicta (Hymenoptera: Formicidae)

Multiple-queen (polygyne) colonies of the introduced fire ant Solenopsis invicta present a paradox for kin selection theory. Egg-laying queens within these societies are, on average, unrelated to one another, and the numbers of queens per colony are high, so that workers appear to raise new sexuals that are no more closely related to them than are random individuals in the population. This paradox could be resolved if workers discriminate between related and unrelated nestmate sexuals in important fitness-related contexts. This study examines the possibility of such nepotism using methods that combine the following features: (1) multiple relevant behavioral assays, (2) colonies with an unmanipulated family structure, (3) multiple genetic markers with no known phenotypic effects, and (4) a statistical technique for distinguishing between nepotism and potentially confounding phenomena. We estimated relatedness between interactants in polygyne S. invicta colonies in two situations, workers tending egg-laying queens and workers feeding maturing winged queens. In neither case did we detect a significant positive value of relatedness that would implicate nepotism. We argue that the non-nepotistic strategies displayed by these ants reflect historical selection pressures experienced by native populations, in which nestmate queens are highly related to one another. The markedly different genetic structure in native populations may favor the operation of stronger higher-level selection that effectively opposes weaker individual-level selection for nepotistic interactions within nests. Received: 28 June 1996 / Accepted after revision: 6 October 1996     

Condition-dependent response to changes in pollen stores by honey bee (Apis mellifera) colonies with different parasitic loads

The impact of a parasitic infestation may be influenced by nutritional state, in both individuals and colonies. This study examined the interaction between pollen storage and the effects of an infestation by the mite, Varroa jacobsoni Oudemans, in colonies of the honey bee, Apis mellifera L. We manipulated the pollen storage and mite infestation levels of colonies, and measured pollen foraging and brood rearing. Increased pollen stores decreased both the number of pollen foragers and pollen load size, while initially at least foragers from colonies with moderate infestations carried smaller pollen loads than those from lightly infested colonies. Over the course of the experiment, all colonies significantly increased pollen-foraging rates and pollen consumption, which was presumably a seasonal effect. Lightly infested colonies exhibited a larger increase in pollen forager number than moderately infested colonies, suggesting that more intense mite infestations compromised forager recruitment. Brood production was not affected by the addition of pollen, but moderately infested colonies were rearing significantly less brood by the end of the experiment than lightly infested colonies. Furthermore, the efficiency with which colonies converted pollen to brood decreased as the pollen storage level decreased and the infestation level increased. The results of this study may indicate that honey bee colonies adaptively alter brood-production efficiency in response to parasitic infestations and seasonal changes. Received: 3 May 1999 / Received in revised form: 14 September 1999 / Accepted: 25 September 1999     

Do these eggs smell funny to you?: an experimental study of egg discrimination by hosts of the social parasite Polyergus breviceps (Hymenoptera: Formicidae)

Social parasites exploit the behaviours of other social species. Infiltration of host systems involves a variety of mechanisms depending on the conditions within the host society and the needs of the social parasite. For many species of socially parasitic ants, colony establishment entails the usurpation of colonies of other species. This frequently involves the eviction or death of the host colony queen and the subsequent adoption of the invading queen. The social parasite queen achieves host worker acceptance by either manipulating the nest-mate recognition processes of the host or undergoing chemical modification. Little is known, however, about how host workers respond to social parasite eggs or whether host species defend against brood parasitism during parasite invasions. Host species are believed to adopt social parasite offspring because the recent common ancestry between many social parasites and their hosts may grant the sharing of certain characteristics such as chemical cues. Use of multiple host species, however, suggests other processes are needed for the social bond between host and parasite young to form. This study reports the findings of adoption bioassays in which eggs from a slave-maker ant, Polyergus breviceps, were offered to workers of two of its host species from unparasitised or newly parasitised nests to determine whether P. breviceps eggs generally elicit rearing behaviours from multiple host species. Comparisons of parasite egg survival until adulthood with conspecific egg survival reveal that workers of both host species, free-living or newly enslaved, do not typically accept slave-maker eggs. Both host species thus have sufficient discriminatory power to reject social parasite eggs although our hydrocarbon analysis indicates parasite eggs may be adapted to their local host species. Combined these results suggest that host rearing of P. breviceps eggs may reflect an evolutionary equilibrium that is maintained by probability and cost of recognition errors.Communicated by L. Sundström     

Almost royal: incomplete suppression of host worker ovarian development by a social parasite wasp

Insect social parasites, like other parasites, may benefit from inhibiting their host from reproducing (complete or partial parasitic castration) because they can then exploit more of the host’s resources for their own reproduction. In particular, social parasites that kill or expel the host queen need to prevent host workers from reproducing; this is a common worker response to the absence of their queen. Indeed, host workers would benefit from detecting the presence of the parasite and investing in direct and indirect fitness. Studying whether and how social parasites control host worker reproduction can provide information about the degree of integration of the parasite in the host colony and help identify factors regulating workers’ reproductive decisions in social insects. We investigated whether the paper wasp social parasite, Polistes sulcifer, suppresses Polistes dominula (host) worker reproduction as efficiently as the dominant host female does in queen-right colonies by comparing worker reproductive efforts in parasitized and non-parasitized (control) colonies. Our results show that 6 weeks after usurpation of their colony by the social parasite, parasitized workers (1) had more developed ovaries than control workers and (2) laid more eggs as soon as the opportunity arose. This reproductive readiness of parasitized workers was not apparent 2 weeks after colony usurpation. This suggests that P. dominula workers have evolved means to react to social parasitism, as occurs in some ants, and that the parasite has only limited control over host reproduction.     

Comparative study of genetic and social structure in two forms of the fire ant Solenopsis invicta (Hymenoptera: Formicidae)

Summary The genetic and social structures of polygyne and monogyne forms of the fire ant, Solenopsis invicta, are investigated in a comparative manner using allozyme data from two polymorphic loci. Foundress queens of the monogyne form are signly inseminated and appear to produce all males present in the colony during the major summer mating flights. The average regression coefficient of relatedness (b) among female nestmates of the monogyne form is 0.714 (Fig. 2), statistically indistinguishable from the pedigree coefficient of relatedness (G) of 0.75. We suggest that the evolution of obligate worker sterility in Solenopsis is associated with this high relatedness between workers and the queens they rear. Functional queens in polygyne nests also are singly inseminated and are no more closely related to nestmate queens than to other queens (within-nest b=0). Within-nest relatedness of workers in the polygyne population is similarly low (Fig. 2). Both the monogyne and polygyne populations from northcentral Georgia are in Hardy-Weinberg equilibrium at both allozyme loci and we found no evidence of significant population subdividion or inbreeding in the polygyne population. These results do not support the view that kin selection has promoted the evolution of polygyny in North American S. invicta. Rather, mutualism appears to be the most likely selective factor mediating queen associations inthis ant.     

Intraspecific brood parasitism in the moorhen: parentage and parasite-host relationships determined by DNA fingerprinting

Parasitic female moorhens (Gallinula chloropus) lay from one to six eggs in the nests of conspecific neighbours. DNA fingerprinting was used to show that parasitic eggs could be correctly identified when they appeared in addition to or outside the host’s laying sequence. Moorhen hosts accept all parasitic eggs laid after the 2nd day of their laying period. To understand why moorhen hosts tolerate parasitic eggs, we tested two hypotheses. (1) The quasi-parasitism hypothesis: females lay their eggs in the evening when the host males are normally in attendance at the nest, so host males may allow parasitic females to lay in their nests in exchange for fertilizing their eggs. However, DNA fingerprinting showed that all the parasitic eggs were sired by the parasites’ mates. Parasitic moorhens frequently continue laying a clutch in their own nest, without a break in the laying sequence after a parasitic laying bout. The eggs laid by brood parasites in their own nests were also sired by their own mates. Therefore this hypothesis was rejected. (2) The kin selection hypothesis: if one or both members of the host pair are close relatives of the parasite, the costs of rearing parasitic chicks will be to some degree offset by inclusive fitness benefits. We examined the genetic relationships between parasites and their hosts using DNA fingerprinting and genealogical data. Natal philopatry by both sexes was relatively common in this population, and the probability that a neighbour of either sex was a first-order relative (parent-offspring) was calculated as 0.18. Although first-order relatives were not preferentially chosen as hosts over individuals that were not first-order relatives, even through random host selection there is almost a one-in-five chance that brood parasites in this population are closely related to their hosts. This may facilitate host tolerance of parasitic eggs. Other hypotheses are also discussed. Received: 3 February 1995/Accepted after revision: 27 August 1995     

Parasitism and ecological relationships among deep-sea benthic fishes

We have studied the metazoan parasite fauna of 52 species of deep-living benthic fishes from depths of 53 to 5000m off the New York Bight (39–49°N; 70–72°W). 17144 parasites were recovered from 1712 fishes. The infestation rate was 80%, with an average of 12.5 worms per host. Percentage occurrence by group among all fishes was Monogenea 12.9%, Digenea 48%, Cestoda 22.1%, Nematoda 54.5%, Acanthocephala 3.8%, and Copepoda 4.5%. Differing composition of the parasite fauna in different fish species reflects differences in diet. Specialized feeders are rather distinct; generalized feeders, which predominate, show overlaps in parasite fauna. In individual species, changes in diet with growth are reflected in changes in the parasite fauna. Infestation rate is directly related to abundance of the free-living fauna; hence, fish from within the submarine canyon are more heavily infested than those living without. Although it contains fewer families and genera than shallow faunas, the deep-sea parasite fauna is not extremely unusual in terms of its abundance, diversity, or host specificity. At the greatest depths, parasite abundance and diversity dramatically decline.     

Density-dependent effects of ants on selection for bumble bee pollination in Polemonium viscosum

Mutualisms are commonly exploited by cheater species that usurp rewards without providing reciprocal benefits. Yet most studies of selection between mutualist partners ignore interactions with third species and consequently overlook the impact of cheaters on evolution in the mutualism. Here, we explicitly investigate how the abundance of nectar-thieving ants (cheaters) influences selection in a pollination mutualism between bumble bees and the alpine skypilot, Polemonium viscosum. As suggested in past work with this species, bumble bees accounted for most of the seed production (78% +/- 6% [mean +/- SE]) in our high tundra study population and, in the absence of ants, exerted strong selection for large flowers. We tested for indirect effects of ant abundance on seed set through bumble bee pollination services (pollen delivery and pollen export) and a direct effect through flower damage. Ants reduced seed set per flower by 20% via flower damage. As ant density increased within experimental patches, the rate of flower damage rose, but pollen delivery and export did not vary significantly, showing that indirect effects of increased cheater abundance on pollinator service are negligible in this system. To address how ants affect selection for plant participation in the pollination mutualism we tested the impact of ant abundance on selection for bumble bee-mediated pollination. Results show that the impact of ants on fitness (seed set) accruing under bumble bee pollination is density dependent in P. viscosum. Selection for bumble bee pollination declined with increasing ant abundance in experimental patches, as predicted if cheaters constrain fitness returns of mutualist partner services. We also examined how ant abundance influences selection on flower size, a key component of plant investment in bumble bee pollination. We predicted that direct effects of ants would constrain bumble bee selection for large flowers. However, selection on flower size was significantly positive over a wide range of ant abundance (20-80% of plants visited by ants daily). Although high cheater abundance reduces the fitness returns of bumble bee pollination, it does not completely eliminate selection for bumble bee attraction in P. viscosum.