Nosema Bombi: a Pollinator Parasite with Detrimental Fitness Effects

Journal of INVERTEBRATE PATHOLOGY Journal of Invertebrate Pathology 96 (2007) 118–124 www.elsevier.com/locate/yjipa

Nosema bombi: A pollinator parasite with detrimental ﬁtness eﬀects

Oliver Otti a,*, Paul Schmid-Hempel a,b

a Experimental Ecology, Institute of Integrative Biology Zurich (IBZ), ETH Zurich, 8092 Zurich, Switzerland b Wissenschaftskolleg zu Berlin, Wallotstrasse 19, D-14193 Berlin, Germany

Received 30 October 2006; accepted 27 March 2007 Available online 31 March 2007

Abstract

Nosema bombi is an obligate intracellular parasite that infects different bumblebee species at a substantial, though variable, rate. To date its pathology and impact on host fitness are not well understood. We performed a laboratory experiment investigating the pathology and fitness effects of this parasite on the bumblebee Bombus terrestris. We experimentally infected one group of colonies with N. bombi spores at the start of the worker production, while a second uninfected group of colonies served as controls. During colony development we collected live workers for dissections to measure infection intensities. In parallel, we measured several life history traits, to investigate costs to the host. We succeeded in infecting 11 of 16 experimental colonies. When infection occurred at an early stage of colony development, virtually all individuals were infected, with spores being found in a number of tissues, and the functional fitness of males and young queens was reduced to zero. Further, the survival of workers from infected colonies and infected males were reduced. With such severe effects, N. bombi appears to decrease its opportunities for transmission to the next host generation. 2007 Elsevier Inc. All rights reserved.

Keywords: Host–parasite interaction; Nosema bombi; Bombus terrestris; Pathology; Fitness eﬀects; Microsporidia; Virulence; Pollinator parasites

1. Introduction could play a role in reducing populations, at least when they invade populations that are diminishing for other Bumblebees are important natural pollinators for a wide reasons. range of flowering plants (Goulson, 2003) and are of eco- Studies on other economically important insect species, nomic importance as they pollinate a substantial propor- e.g. the silkworm moth and the honeybee, have shown that tion of our crops. They are therefore commercially bred parasites can inflict substantial financial losses to managed in large numbers to be used in the pollination of a range populations. For example, in the 19th century, the micro- of greenhouse crops, such as tomato, capsicum and melon sporidian parasite N. bombycis, the agent of pebrine disease (Al-Attal et al., 2003; Banda and Paxton, 1991; de Ruijter, in the silkworm Bombyx mori (an insect important for the 1997). In the United States alone, the annual economic production of textiles), was spread throughout France and value of pollination services is estimated at 1.25 billion caused massive losses to the silk industry (Cadeddu, 2000). USD, of which bumblebees provide a large share (Ghazoul, Currently, the Varroa mite in honeybees poses financial 2005). These major ecological and economic roles have led problems to the apiculture industry as it impairs the honey to widespread concern over recent declines in pollinator production and kills off reproductive animals so that the populations, including bumblebees (Goulson, 2003; Wil- colonies can no longer be propagated (Sammataro et al., liams, 2003). The major causes of this decline are thought 2000). In the United States, the industry is under additional to relate to changes in the use of agricultural land (Goulson pressure due to the spread of the Africanized honeybee et al., 2005; Williams, 2005), but also the effects of parasites (Whitfield et al., 2006). There is, however, a concern that a parasite, the microsporidian Nosema bombi Fantham and Porter, 1914, could cause the decline of the native * Corresponding author. Fax: +41 44 632 1271. bumblebee community (Kearns and Thomson, 2001; Whit- E-mail address: [email protected] (O. Otti). tington and Winston, 2003), but too little is known to date

0022-2011/$ - see front matter 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.jip.2007.03.016 O. Otti, P. Schmid-Hempel / Journal of Invertebrate Pathology 96 (2007) 118–124 119 about the interactions of N. bombi with its bumblebee host, 2. Materials and methods e.g. impact on host fitness and the transmission dynamics (Fries et al., 2001), to make any informed estimate of the 2.1. Preparation of inoculum threat posed by this microsporidium. The microsporidians are related to fungi (Adl et al., To prepare the inoculum, abdomens of infected bumble- 2005; James et al., 2006), and all the members of this phy- bees from four different colonies (50 workers per colony) lum are obligatory intracellular parasites. The host range were homogenised in Krebs Ringer solution. The homoge- includes nearly all animal groups from protists to humans, nate was filtered through glass wool. The filtered spore solu- with insects as one of the major host groups (Cali and tion was centrifuged for 5 min at 3000g and 4 Cand Takvorian, 1999). In general, microsporidians have two checked for N. bombi using a light microscope. Ammonium major life cycle stages, a spore stage and a vegetative stage. chloride, which inhibits the germination process, was added The infective or environmental stage is the spore, and in to the solution to minimize the loss of infective spores (Und- most cases is ingested by the host and first infects host cells een and Avery, 1988). Subsequently, the spores were washed in the gut lumen and the Malpighian tubules. During and centrifuged another two times. After washing, the con- reproduction and proliferation the parasite spreads within centration of spores was determined in a haemocytometer the host. Fresh spores are then released into the environ- (Neubauer improved). Aliquots were prepared (100 llwith ment via faeces or a decaying host (Cali and Takvorian, a concentration of 5.5 · 105 spores llÀ1)andkeptat 1999). The spore has a thick chitinous cell wall that pro- À80 C until needed for the experiment. tects it from harsh environmental conditions, i.e. temperature and humidity (Li et al., 2003; Maddox and Solter, 2.2. Exposure to N. bombi (treatments) 1996; Malone et al., 2001). Bombus terrestris and other Bombus species are infected We selected 94 B. terrestris queens that originated from by N. bombi at substantial though variable rates (Fries first generation laboratory-reared colonies and were mated et al., 2001; Imhoof and Schmid-Hempel, 1999; Schmid- randomly with males from other laboratory colonies. After Hempel and Loosli, 1998; Shykoff and Schmid-Hempel, artificial hibernation the queens were allowed to oviposit in 1991; Tay et al., 2005). N. bombi primarily infects the Mal- the laboratory under standard conditions (Gerloff and Sch- pighian tubules, but also infects the thorax muscles, fat mid-Hempel, 2005). After the first egg batch was laid, the body tissue, nerve tissue, midgut and the muscle tissue queens were assigned at random to one of two treatments. surrounding the gut epithelium (Fries et al., 2001). In the Queens in the ‘‘Exposed’’ treatment received a pollen pellet Malpighian tubules spores and vegetative stages are found with 18 ll of spore solution (5.5 · 105 spores llÀ1, or a total in the regenerative cell layer of the gut epithelium near the of 9.9 · 106 spores per feeding) twice a week for 4 weeks. basement membrane (McIvor and Malone, 1995). In heavy Queens in the ‘‘Control’’ treatment were fed pollen pellets infections, the structure of the tissues is destroyed and with only the suspension solution. Because imagos cannot mature spores are released into the lumen. The queen is be reliably infected with N. bombi (Otti, unpublished data) the primary source of N. bombi infection for the next our method mimics a natural exposure of larvae to the para- generation, and infection has seemingly little impact on sitevia contact withthe nursing queen. To reduce the potential her nest initiation and egg-laying capacity (Fisher and for inflated re-infection rates due to the housing conditions of Pomeroy, 1989b). colonies in the laboratory, dead workers were removed every Data on the effect of N. bombi on the fitness of its second day. Separate feeding and defecating boxes with cat lit- bumblebee host is scarce and contradicting. Some authors ter substrate were exchanged weekly, and sugar water feeders have reported chronic rather than fatal effects (Fisher and were cleaned and filled with fresh sugar water every week. Pomeroy, 1989a; Imhoof and Schmid-Hempel, 1999), while During the treatment period 18 queens died (11 controls, 7 others found more severe effects (DeJonghe, 1986; Fantham exposed), 43 queens either stopped egg-laying or their brood and Porter, 1914; Macfarlane et al., 1995; Schmid-Hempel did not develop (20 controls, 23 exposed); from 30 broods at and Loosli, 1998). Most of these reports are based on either least one worker emerged (14 controls, 16 exposed). Three accidental field observations or unintended (natural occur- queens were excluded from the experiment as they died due ring) infections in laboratory experiments. Data from to a malfunction of the feeding equipment. As soon as the field-collected live bumblebees that are dissected and colony size reached 10 workers we culled each week 20% of assessed for parasite infestation might give a biased picture the workers to mimic a natural mortality rate. The culled of the overall effect. For example, parasites inflicting heavy workers were used to check for infection status, and these mortality may go undetected, and prevalence and effects of revealed that 11 of the 16 exposed colonies had at least one disease underestimated. We, therefore, conducted an exper- infected individual (infection success 68.75%). iment under standardised laboratory conditions to investigate the impact and pathology of N. bombi in its 2.3. Colony level measures bumblebee host B. terrestris. We tested the effect of experimental infections on a number of life history, pathology During colony development we measured the time until and fitness parameters. the adult eclosion of the first worker, male and gyne 120 O. Otti, P. Schmid-Hempel / Journal of Invertebrate Pathology 96 (2007) 118–124

(unmated queen). As a measure for the development of the number per image was counted. We averaged counts over different castes we used the time in days from first egg-lay- the 10 images. The same males were checked for infection, ing to the eclosion of the first individual of each caste. We and if we found an infection we measured the infection counted the numbers of individuals produced in each caste, intensity in the Malpighian tubules with a haemocytometer and recorded the survival of the maternal queen. (Neubauer improved). The Malpighian tubules were separated from the ventriculus and rectum and gently washed 2.4. Individual level measures in double-distilled water. Then, we transferred the tubules into a 1.5 ml Eppendorf tube with 100 ll double-distilled 2.4.1. Workers water, homogenised and vortexed them before counting. We counted the number of dead workers removed from Four control colonies and six experimental colonies pro- each colony, giving an estimate of the effect of N. bombi on duced at least one young queen (gyne). The freshly emerged worker survival. For analysis we calculated the proportion gynes were put in plastic boxes of up to eight individuals. of dead workers per colony. Every week we took faeces In only two colonies did we find the gynes to be infected, samples from at least two workers per colony to check and one of these colonies produced only six gynes. In total, for infection. Faeces samples are indicative of an individ- we dissected 36 gynes to elucidate the location of infection. ual’s infection status (infected/not infected) and faecal sam- Additionally, we conducted mating trials pairing infected ples typically correlate with within-host intensity (at least gynes (N = 16) with uninfected industrially bred males. for C. bombi)(Otterstatter and Thomson, 2006). However, Although, enough infected gynes were produced to con- we here were also interested in patent infections, i.e. those duct the mating trials, 25% of them died before the day that can be transmitted to the next host, which requires of the mating trial; if the gynes survived they would not cells to appear in the faeces. We analysed a total of 568 fae- let the males copulate. ces samples. Additionally, to determine the tissues infected by the parasite, we dissected two workers from each 2.5. Pathology of the parasite colony. By dissecting bumblebee workers, queens and males we 2.4.2. Sexuals gained information about the location and intensity of For each dissected individual we assessed the body infection. The tissues of main interest were the gut, Malpi- weight to the nearest milligram and body size. Body size ghian tubules and fat body in imagos, accessory glands, was quantified by measuring the length of the radial cell ovaries in gynes, and testicular organs in males. on the right wing to the nearest 0.001 mm using a digital imaging system and the public domain ImageJ program 2.6. Statistical analysis (National Institutes of Health, USA). This measure is strongly correlated with body size in general (Owen, As some of the exposed colonies were not infected we 1988; Owen, 1989). Nine control colonies and 13 experi- performed all the statistical tests as follows. First, we inves- mental colonies produced at least one male. Four of the tigated the effect of the treatment (two-level test—Exposed 13 experimental colonies had at least one infected male. vs. Control). Second, we further separated the treatment After the first males emerged we placed them in plastic ‘‘Exposed’’ into two categories—‘‘Infected’’ and ‘‘Unin- boxes with up to 10 individuals. We counted the number fected exposed’’—giving three levels (three-level test). In of males that died in these boxes over a period of three the result section we only report results of the three-level weeks, i.e. measuring the male mortality. Additionally, test, if we also had found a significant difference in the we dissected 10–15 day old males (N = 150) to locate the two-level test. If the actual treatment effect (two levels) infection, in order to determine the infection intensity in was not significant, neither was the three-level test. ANO- the Malpighian tubules and the viability and number of VAs were conducted, using the shareware statistical soft- sperm. Males were dissected in Kiev:M saline (0.3 g Glu- ware R1.0.8 (Crawley, 2002). All transformations of cose, 0.41 g KCl, 0.21 NaHCO3, 2.43 g C6H5Na3O7Æ2H2O; response variables were calculated with the Box-Cox func- 1000 ml ddH2O, pH 8.8). The content of one seminal vesi- tion, which calculates in an iterative process an exponential cle was gently diluted in 8 ll Kiev:M saline on a micro- transformation variable lambda, k. ANOVA was used to scope slide. Sperm viability assessment and sperm counts find significant effects of treatment on the developmental were carried out with the LIVE/DEAD Sperm Viability time of different castes, the number of workers produced, Kit (L-7011, Molecular Probes). We recorded 10 images the queen survival, the proportion of dead workers col- in set locations on each slide with a fluorescence micro- lected per colony, and the sperm number of males. For scope (Nikon Eclipse with the following filters the numbers of males and gynes produced we conducted 420 6 EX 6 490, DM = 510, BA = 520; Nikon Corpora- ANCOVA with the number of workers produced as a tion, Tokyo, Japan) and a digital camera (Nikon Coolpix covariate, because the colony size had a significant effect 990, Nikon Corporation, Tokyo, Japan). From each on the number of sexual offspring produced. Within the sample 100 sperm, 10 on each of the 10 sub-samples were infected individuals we tested also whether spore load of scored as red (dead) or green (live), and the total sperm worker faeces and the infection intensity in male O. Otti, P. Schmid-Hempel / Journal of Invertebrate Pathology 96 (2007) 118–124 121

Malpighian tubules was different between the colonies, 1.0 using ANOVA. Binary logistic regressions were performed with SPSS 11.0 for Mac OSX to test for treatment effects 0.9 on colony foundation, on whether or not males or gynes were produced, and on male survival. For the male survival 0.8 the first thirty males produced from each colony were ana- 0.7 lysed for a treatment effect on survival. 0.6 3. Results 0.5 3.1. Colony level measures 0.4 The proportion of queens that founded a colony (i.e. Proportion of dead workers 0.3 produced workers), then produced males or gynes did not differ between treatments (Binary logistic regression; for 0.2 colony founding: Wald1,91 = 0.139, P = 0.71, males: 0.1 Waldz1,91 = 1.296, P = 0.26, gynes: Wald1,91 = 0.397, P = 0.53). When looking at the developmental times of 0.0 the different castes (days from egg-laying to hatching, Control (13) Infected (10) Uninfected workers, males, gynes) we also did not find any difference exposed (5) between controls and exposed groups (ANOVA; workers: Infection status of colony F = 0.664, P = 0.42, males: F = 0.568, P = 0.46, 1,29 1,22 Fig. 1. Mean proportion of dead workers removed from control (white gynes: F1,10 = 0.871, P = 0.38). Further, the number of bar), infected (black bar) and uninfected exposed (grey bar). Error bars workers, males and gynes produced per colony did not dif- represent 95% CI for binomial distributions. Numbers in brackets refer to fer between the control and the infected colonies (ANOVA; the number of colonies in each category. workers: k = 0.2, F1,28 = 1.136, P = 0.3; ANCOVAs with colony size as covariate, males: k = 0.54, F1,22 = 2.64, P = 0.12, gynes: k = À0.14, F1,10 = 0.048, P = 0.83). 3.5 Founding queen survival (from colony start to death of

the queen) was also not diﬀerent between the two treat- ) 5 3.0 ments (ANOVA; F1,38 = 0.0003, P = 0.99). l (x 10 3.2. Individual level measures µ 2.5

3.2.1. Workers 2.0 The mortality of infected workers was significantly higher than that of uninfected workers (ANOVA: 1.5 F1,28 = 13.52, P = 0.001). The model for the three-level test was also significant (Fig. 1; ANOVA: F2,28 = 7.224, P = 0.003). In post hoc pair-wise tests, the ‘‘Infected’’ group 1.0 was significantly different from ‘‘Control’’, but not from

the ‘‘Uninfected exposed’’ (pairwise t-test with Bonferron- Mean spore count in spores/ 0.5 i-adjusted P-value: P = 0.003 and P = 0.23, respectively). The ‘‘Control’’ group did not differ from the ‘‘Uninfected exposed’’ (pairwise t-test with Bonferroni-adjusted P-value: 0 31 53 12 11 P = 0.84). The spore content in the worker faeces was Colony found to be variable among infected colonies (Fig. 2; Fig. 2. Mean spore content of worker faeces in spores llÀ1 of faeces per ANOVA: F4,106 = 8.425, P < 0.001). On average, 48.6% of the sampled workers had spores in their faeces. The infected colony. The numbers below bars are number of worker faeces samples per colony. Error bars represent one standard error. median spore content was 14,221 spores llÀ1 of faeces (N = 106, Inter-quartile range, IQR = 77,104À1588 spores llÀ1). The three-level test revealed that survival was significantly 3.2.2. Sexuals lower for infected colonies than for control and uninfected Male survival over 21 days was significantly lower for colonies (Fig. 3; Binary logistic regression: Wald8,390 = those emerging from colonies in the exposed treatment 20.44, P < 0.001). For the dissected males the body weight (Binary logistic regression: Wald6,390 = 10.925, P =0.001). and size did not significantly differ between control and 122 O. Otti, P. Schmid-Hempel / Journal of Invertebrate Pathology 96 (2007) 118–124

1.0 30

0.9 25 0.8

0.7 20 0.6

0.5 15

0.4 10 0.3 Mean number of sperm 0.2 5 0.1

Proportion of individuals surviving 21 days after emergence 0.0 0 Control (180) Infected (120) Uninfected Control (67) Infected (49) Uninfected exposed (90) exposed (33) Infection status of colony Infection status of colony Fig. 3. Proportion of males surviving 21 days after emergence shown for Fig. 4. Mean sperm number in males from control (white bar), infected control (white bar), infected (black bar) and uninfected exposed colonies (black bar) and uninfected exposed colonies (grey bar). Error bars (grey bar). Sample sizes given in brackets. Error bars represent 95% CI for represent one standard error. Sample sizes given in brackets. binomial distributions.

4.5 exposed colonies (ANOVA, body weight: F1,150 = 1.26, P = 0.26; body size: F1,147 = 0.181, P = 0.67).

) 4.0

The mean sperm number in males from exposed colonies 5 was signiﬁcantly lower than the sperm number in controls 3.5 (ANOVA: F = 10.24, P = 0.002). The three-level test 1,148 l (x 10

µ revealed that the males from infected colonies had a signif- 3.0 icantly lower number of sperm than males from the control and uninfected exposed colonies (Fig. 4; pairwise t-test 2.5 with Bonferroni-adjusted P-value: P < 0.001, respectively, P < 0.001). 2.0 The variation in infection intensity in the Malpighian tubules in males from infected colonies paralleled the vari- 1.5 ation found in the worker’s faeces spore loads. Overall, the intensities diﬀered signiﬁcantly between colonies (Fig. 5; 1.0 ANCOVA: F5,30 = 10.862, P < 0.001). The median spore Mean spore count in spores/ content of male Malpighian tubules was 215,000 spores 0.5 llÀ1 (N = 30, IQR = 315,000À111,250 spores llÀ1). After dissection infected males can be distinguished from unin- 0 102510 3 fected ones by swollen and shiny white Malpighian tubules. Also, the testes and accessory testes loose their natural Colony structure. When attempting to quantify the number of Fig. 5. Mean malpighian infection intensity of males in spores llÀ1 per sperm cells in male accessory testes, we discovered that infected colony. The numbers below bars are number of males per colony. these organs were full of Nosema spores in the infected Error bars represent one standard error. males; also their testes contained spores. In one of the two infected colonies that produced infected gynes, 95.71% of those showed distended abdo- We found spores in the gut, Malpighian tubules and fat mens or crippled wings. We were unsuccessful in mating body of all infected individuals. Furthermore, spores of these crippled gynes with healthy males. And 25% of the N. bombi were discovered in the accessory glands and crippled gynes did not even survive until day 6 following ovaries of infected gynes and in the testicular organs of eclosion (the age at which matings took place). infected males. O. Otti, P. Schmid-Hempel / Journal of Invertebrate Pathology 96 (2007) 118–124 123

4. Discussion reproduction, as earlier studies showed that the colony size is an important factor determining the probability ‘‘Classical’’ fitness parameters, such as colony size and and the number of sexuals produced (Mu¨ller and Sch- number of sexuals produced, were statistically not influ- mid-Hempel, 1992). In addition to mortality, N. bombi enced by the treatment. However, when considering the seems to have an effect on the behaviour of the workers. performance of the produced sexuals, that is, the chances We did not systematically measure these components, of daughter queens and males to have offspring themselves, but several characteristics were observed in infected colo- the parasite showed strong fitness effects. nies and individuals. For example, an infected colony Infected males have lower survival and almost no appears dirty. This could be due to the diarrhoea caused sperm, which decreases their actual fitness drastically. In by the parasite and the inefficient cleaning behaviour of addition, the lower survival in males likely reduces the the workers, as it is the case in infected honeybees (Fries, number of potential mates that each female can encounter 1993). In heavily infected colonies dead individuals are and, thus, the number of successful matings. The observa- not carried outside the nest and the live workers appear tion of parallel variation in spore content in worker faeces slow and clumsy. These observations indicate that work- and infection intensity in the male Malpighian tubules ers of infected colonies are no longer able to efficiently among colonies hints to possible genotype–genotype fulfil their job. interactions underlying the relationship between the par- The pathogenicity of N. bombi seems an evolutionary asite and the host (Schmid-Hempel and Loosli, 1998). puzzle. By crippling the only carriers (the young gynes) On the other hand, genetic studies on N. bombi have that last through the hibernation period (as far as it is failed to reveal a large degree of variation among isolates known, N. bombi cannot survive winter outside its host) (Tay et al., 2005). Therefore, although our results suggest N. bombi reduces its chances to get into the next host a variable genotypic component in the host, we can only generation. We have no readyexplanationforthese state that there is potential for these interactions to occur. effects. In the laboratory the unlimited access to food The infected gynes (only two colonies produced visibly might allow higher infection intensities, and workers live infected individuals) were impaired in their behaviour longer than in the field. Due to the confinement of due to the swelling of their abdomens. Further, a large pro- healthy and diseased bees re-infection rates might be portion of the gynes in one colony had crippled wings and higher as well. Nevertheless, in the field the first worker so could not fly. Recently, WDV (wing-deformed virus) brood will quickly build up high infection intensities as has been documented in bumblebees (Genersch et al., they are all fed by the infected mother queen, thus being 2006) and it is possible that this virus could have caused repetitively and continuously exposed to N. bombi the crippled wings (the presence of viruses could not be spores. In the field workers face more stressful condi- checked in our study). However, it would then remain tions and probably die at lower infection intensities than unclear why the effects of the virus are only expressed in in the laboratory, therefore reducing the colony size. The N. bombi-infected colonies. Even though gynes are pro- smaller the colony size, the lower are the chances for the duced in infected colonies, their reproductive potential is colony to reach the stage of sexual reproduction (Imhoof virtually non-existent. The mating trials showed that infec- and Schmid-Hempel, 1999; Michener, 1964; Mu¨ller and tion by the parasite reduces the readiness of the gynes to Schmid-Hempel, 1992). It is therefore likely that our mate, as has been previously reported (Macfarlane et al., lab-based results are conservative. In harsher field condi- 1995). The reproductive’s of both sexes therefore showed tions, infected colonies may have a greatly reduced severe effects of infection, potentially reducing their func- probability of reaching the stage of sexual reproduction, tional fitness to zero. As some of the sexual individuals as the workers die off early in the field. Therefore, under of an infected colony do not become infected—at least natural conditions, the virulence of N. bombi may pose a according to visible inspection—an infected colony can threat to itself. also produce healthy sexuals (we estimated 35.1% of 130 Our study only considers the impact of Nosema on checked sexuals) that will mate and go into hibernation. early-infected colonies, and it is possible that in the field N. bombi, like many other microsporida, can cause chronic only these will be impaired to the observed extent. Infec- infections that may be difficult to detect and may not tions that happen at a later stage of colony development always lead to severe pathologies. could be less severe and still allow the colony to produce Whereas the total number of workers produced was viable gynes that act as carriers. Late stage exposure might not affected by the parasite, the proportion of workers generate low-level infections that are extremely difficult to that died during the experiment was significantly higher observe, and may be one mechanism that would allow in infected colonies than in the controls. Under stressful infected females overwintering and thus to transmit the natural conditions this is expected to lead to either smal- parasite to the next generation. Clearly, the pathology, as ler colony size or the need for increased production rates well as modes and potential for transmission of N. bombi, of new workers in infected colonies, but this has so far not still need further elucidation and, therefore, further inves- been documented (Imhoof and Schmid-Hempel, 1999). tigation on the ecology of this host–parasite interaction Such a scenario could have an impact on the colony’s is needed to answer these questions. 124 O. Otti, P. Schmid-Hempel / Journal of Invertebrate Pathology 96 (2007) 118–124

Acknowledgments Ghazoul, J., 2005. Buzziness as usual? Questioning the global pollination crisis. Trends in Ecology and Evolution 20, 367–373. We thank L. Wilfert, M. Greeff and C.P. Yourth for help- Goulson, D., 2003. Bumblebees: Their Behaviour and Ecology. Oxford University Press, Oxford. ing with methods and bee care, N. Rossel and P. Nigg for Goulson, D. et al., 2005. Causes of rarity in bumblebees. Biological phenotypic measurements. We also thank B. Sadd, R. Pax- Conservation 122, 1–8. ton, J. Klee and I. Fries for discussion and two anonymous Imhoof, B., Schmid-Hempel, P., 1999. Colony success of the bumble reviewers for helpful comments on our manuscript. Finan- bee, Bombus terrestris, in relation to infections by two protozoan cially supported by grants to PSH from the Swiss National parasites, Crithidia bombi and Nosema bombi. Insectes Sociaux 46, 233–238. Science Foundation (No. 0031-66733.01) and the European James, T.Y. et al., 2006. Reconstructing the early evolution of fungi using Union (HPRN-CT-2000-00052 /BBW 99.0106-2). a six-gene phylogeny. Nature 443, 818–822. Kearns, C.A., Thomson, J.D., 2001. The natural history of bumblebees: a sourcebook for investigations. University of Colorado Press, Boulder, References CO. Li, X. et al., 2003. Infectivity of microsporidia spores stored in water at Adl, S.M. et al., 2005. The new higher level classification of eukaryotes environmental temperatures. Journal of Parasitology 89, 185–188. with emphasis on the taxonomy of protists. The Journal of Eukaryotic Macfarlane, R.P. et al., 1995. Bumble bee pathogens and internal Microbiology 52, 399–451. enemies. Bee World 76, 130–148. Al-Attal, Y.Z. et al., 2003. Influence of pollination technique on Maddox, J.V., Solter, L.F., 1996. Long-term storage of infective micro- greenhouse tomato production. Agricultural and Marine Sciences 8, sporidian spores in liquid nitrogen. Journal of Eukaryotic Microbiol- 21–26. ogy 43, 221–225. Banda, H.J., Paxton, R.J., 1991. Pollination of greenhouse tomatoes by Malone, L.A. et al., 2001. Effects of time, temperature, and honey on bees. Acta Horticulturae 288, 194–198. Nosema apis (Microsporidia: Nosematidae), a parasite of the honey- Cadeddu, A., 2000. The heuristic function of ‘error’ in the scientific bee, Apis mellifera (Hymenoptera: Apidae). Journal of Invertebrate methodology of Louis Pasteur: the case of the silkworm diseases. Pathology 77, 258–268. History and Philosophy of the Life Sciences 22, 3–28. McIvor, C.A., Malone, L.A., 1995. Nosema bombi, a microsporidian Cali, A., Takvorian, P.M., 1999. Developmental morphology and life pathogen of the bumble bee Bombus terrestris (L). New Zealand cycles of the Microsporidia. In: Wittner, M., Weiss, L.M. (Eds.), The Journal of Zoology 22, 25–31. Microsporidia and Microsporidiosis. American Society for Microbi- Michener, C.D., 1964. Reproductive efficiency in relation to colony size in ology, Washington, DC, pp. 85–128. hymenopterous societies. Insectes Sociaux V11, 317–341. Crawley, M.J., 2002. Statistical Computing. An Introduction to Data Mu¨ller, C.B., Schmid-Hempel, P., 1992. Correlates of reproductive success Analysis Using S-plus. John Wiley & Sons, Chichester. among field colonies of Bombus lucorum—the importance of growth de Ruijter, A., Commercial bumble bee rearing and its implications. in: and parasites. Ecological Entomology 17, 343–353. Richards, K.W. (Ed.), Proceedings of the 7th International Sympo- Otterstatter, M.C., Thomson, J.D., 2006. Within-host dynamics of an sium on Pollination, vol. 437. Acta Horticulturae, Lethbridge, Alberta, intestinal pathogen of bumble bees. Parasitology 133, 749–761. Canada, 1997, pp. 261–269. Owen, R.E., 1988. Body size variation and optimal body size of bumble DeJonghe, R., 1986. Crossing experiments with Bombus terrestris terrestris bee queens (Hymenoptera: Apidae). Canadian Entomologist 120, 19– (Linnaeus, 1758) and Bombus terrestris xanthopus (Kriechbaumer, 27. 1870) and some notes on diapause and nosemose (Nosema) (Hyme- Owen, R.E., 1989. Differential size variation of male and female noptera, Apoidea). Phegea 14, 19–23. bumblebees. Journal of Heredity 80, 39–43. Fantham, H.B., Porter, A., 1914. The morphology, biology and economic Sammataro, D. et al., 2000. Parasitic mites of honey bees: Life History, importance of Nosema bombi, N. sp., parasitic in various humble bees implications, and impact. Annual Review of Entomology 45, 519–548. (Bombus spp.). Annuals of Tropical Medicine and Parasitology 8, 623– Schmid-Hempel, P., Loosli, R., 1998. A contribution to the knowledge of 638. Nosema infections in bumble bees, Bombus spp. Apidologie 29, 525– Fisher, R.M., Pomeroy, N., 1989a. Incipient colony manipulation, 535. Nosema incidence and colony productivity of the bumble bee Bombus Shykoff, J.A., Schmid-Hempel, P., 1991. Incidence and effects of 4 terrestris (Hymenoptera, Apidae). Journal of the Kansas Entomolog- parasites in natural populations of bumble bees in Switzerland. ical Society 62, 581–589. Apidologie 22, 117–125. Fisher, R.M., Pomeroy, N., 1989b. Pollination of greenhouse muskmelons Tay, W.T. et al., 2005. Complete rRNA gene sequences reveal that the by bumble bees (Hymenoptera, Apidae). Journal of Economic microsporidium Nosema bombi infects diverse bumblebee (Bombus Entomology 82, 1061–1066. spp.) hosts and contains multiple polymorphic sites. The Journal of Fries, I., 1993. Nosema apis—a parasite in the honey bee colony. Bee Eukaryotic Microbiology 52, 505–513. World 74, 5–19. Undeen, A.H., Avery, S.W., 1988. Effect of anions on the germination of Fries, I. et al., 2001. Molecular characterization of Nosema bombi Nosema algerae (Microspora: Nosematidae) spores. Journal of Inver- (Microsporidia: Nosematidae) and a note on its sites of infection in tebrate Pathology 52, 84–89. Bombus terrestris (Hymenoptera: Apoidea). Journal of Apicultural Whitfield, C.W. et al., 2006. Thrice out of Africa: Ancient and recent Research 40, 91–96. expansions of the honey bee, Apis mellifera. Science 314, 642–645. Genersch, E. et al., 2006. Detection of deformed wing virus, a honey bee Whittington, R., Winston, M.L., 2003. Effects of Nosema bombi and its viral pathogen, in bumble bees (Bombus terrestris and Bombus treatment fumagillin on bumble bee (Bombus occidentalis) colonies. pascuorum) with wing deformities. Journal of Invertebrate Pathology Journal of Invertebrate Pathology 84, 54–58. 91, 61–63. Williams, N., 2003. Plight of the bumblebee. Current Biology 13, R463. Gerloff, C.U., Schmid-Hempel, P., 2005. Inbreeding depression and family Williams, P., 2005. Does specialization explain rarity and decline among variation in a social insect, Bombus terrestris (Hymenoptera: Apidae). British bumblebees? A response to Goulson et al. Biological Conser- Oikos 111, 67–80. vation 122, 33–43.