Transmission of a bumblebee parasite is robust despite parasite exposure to extreme temperatures

Abstract All organisms are exposed to fluctuating environmental conditions, such as temperature. How individuals respond to temperature affects their interactions with one another. Changes to the interaction between parasites and their hosts can have a large effect on disease dynamics. The gut parasite, Crithidia bombi, can be highly prevalent in the bumblebee, Bombus terrestris, and is an established epidemiological model. The parasite is transmitted between bumblebees via flowers, exposing it to a range of environmental temperatures prior to infection. We investigated whether incubation duration and temperature exposure, prior to infection, affects parasite infectivity. Prior to inoculation in B. terrestris, C. bombi was incubated at 10, 20, 30, 40 or 50°C for either 10 or 60 min. These times were chosen to reflect the length of time that the parasite remains infective when outside the host and the rate of floral visitation in bumblebees. Prevalence and infection intensity were measured in bees 1 week later. Incubation duration and the interaction between incubation temperature and duration affected the prevalence of C. bombi at 50°C, resulting in no infections after 60 min. Below 50°C, C. bombi prevalence was not affected by incubation temperature or duration. Extreme temperatures induced morphological changes in C. bombi cells; however, infection intensity was not affected by incubation duration or temperature. These results highlight that this parasite is robust to a wide range of temperatures. The parasite was not infective after being exposed to 50°C for 60 min, such temperatures likely exceed the flight abilities of bumblebees, and thus the potential for transmission. This study shows the importance of understanding the effects of environmental conditions on both hosts and parasites, which is needed to predict transmission under different environmental conditions.


| INTRODUC TI ON
To survive, all organisms must respond to changing environments.
One variable environmental condition is temperature, which fluctuates substantially over a wide range of scales. For example, temperature can vary daily, seasonally, and across the year. In addition, mean temperatures are predicted to increase due to climate change in the coming decades (Wuebbles et al., 2017), posing a further challenge to organisms. All living organisms are sensitive to changes in temperature, and how individuals respond to these changes may affect their interactions with other species (Boukal et al., 2019;Doney et al., 2011). For host-parasite interactions, temperature can have large impact on disease dynamics (Altizer & Oberhauser, 1999;Frid & Myers, 2002;Menti et al., 2000). Consequently, understanding how temperature affects these interactions is necessary for predicting disease dynamics across seasons, in different climates and with climate change.
Temperature can affect disease dynamics directly through altering host susceptibility to infection (Adamo & Lovett, 2011;Murdock et al., 2012). Less obviously, but arguably more importantly, temperature can also affect host susceptibility indirectly. Higher temperatures may result in increased host growth and in turn, host density, leading to higher rates of transmission (Burdon & Chilvers, 1982).
Parasites can be exposed to varying temperatures both inside and outside a host, and these temperatures can affect their survival rate, development rate (Kalinda et al., 2017;Leathwick, 2013) and infectivity (O'Connor et al., 2006). Within a host, parasites that infect poikilotherms, such as arthropods, are exposed to wider fluctuations in temperature compared to parasites of homeothermic hosts. Such fluctuations have major impacts on many vector-borne diseases, such as those transmitted by mosquitoes (Liu-Helmersson et al., 2014). In bumblebees and honeybees, Temperature variation has been shown to correlate with parasite prevalence and infection intensity (Chen et al., 2012;Manlik et al., 2023;McMullan & Brown, 2005;Retschnig et al., 2017). For example, Nosema cerenae infection in honeybee workers has been shown to correlate negatively with temperature (Chen et al., 2012;Retschnig et al., 2017), but this relationship can change depending on other factors, such as seasonality and host genotype (Manlik et al., 2023). Similar relationships exist outside of invertebrate hosts, for example, amphibian chytrid fungus that generally exhibits higher prevalence and infection intensity at lower temperatures (Ellison et al., 2020;Raffel et al., 2015;Sonn et al., 2017).
When outside the host, parasites may be exposed to highly variable environmental conditions. Therefore, parasites with free-living stages and indirect transmission modes are particularly vulnerable to environmental temperature changes due to their period in the external environment. For example, the survival and infectivity of free-living nematode larvae of sheep are affected by environmental temperature, with the optimum temperature varying between species (Morgan & van Dijk, 2012;O'Connor et al., 2006). Temperature has also been shown to affect the survival and viability of viruses in air-borne droplets (Chen, 2020;Prussin et al., 2018), faeces (Moe & Shirley, 1982) and in water sources (Nasser & Oman, 1999). The direction of temperature effects on viral survival and infectivity can vary between viruses and with other environmental conditions, such as humidity (Chen, 2020;Moe & Shirley, 1982;Prussin et al., 2018).

Many pollinator parasites, including Vairimorpha apis, V. bombi,
and Crithidia bombi are transmitted indirectly via the shared use of flowers (Adler et al., 2018;Durrer & Schmid-Hempel, 1994;Figueroa et al., 2019;Graystock et al., 2015;Pinilla-Gallego et al., 2022). These parasites are deposited across the whole flower, for example on the petals and in the nectar, by workers who may defecate when they forage (Durrer & Schmid-Hempel, 1994;Figueroa et al., 2019;Pinilla-Gallego et al., 2022). The subsequent location of the parasite on the flower will determine its exposure to the ambient temperature. Parasite cells on the petals will exhibit higher temperature variation compared to those in the nectar due to both exposure to direct sunlight and the higher specific heat capacity of nectar, a liquid, compared to the petals, a solid. As one would expect, parasite survival can be higher when deposited inside the floral structure, in the corolla, compared on more exposed structures, such as the bract (Figueroa et al., 2019).
Furthermore, the duration of time the parasite is exposed to the environment depends on the visitation rate of the flower by pollinators.
Visitation rate varies between flowers and correlates with the rate of nectar replenishment (Stout et al., 1998). In some cases, the period between flower visits may be up to an hour (Stout & Goulson, 2001).
Interestingly, the temperature of peak growth in vitro does not align with the temperature at which hosts exhibit peak infection intensi- Given the changes in C. bombi growth in vitro at different temperatures (Palmer-Young et al., 2018, it is plausible that temperature exposure on flowers affects its transmission between colonies. Furthermore, temperature may affect within colony transmission of the parasite. Although bumblebees try to maintain a constant temperature within the colony, through fanning when it is hot (Weidenmüller et al., 2002) and incubating (O'Donnell & Foster, 2001) and building wax coverings when it is cold (Jones & Oldroyd, 2006), in reality, the temperature of the colony will fluctuate with environmental temperature and colony size (Crall et al., 2018;Vogt, 1986).
In addition, colony temperature varies between colonies in different climates, particularly at the start of their lifecycle when the colony temperature is similar to the ambient temperature (Hasselrot, 1960).
To test whether temperature exposure affects C. bombi infectivity, we exposed C. bombi to 10, 20, 30, 40 and 50°C for either 10 or 60 min in vitro and inoculated Bombus terrestris audax hosts. One week later, host prevalence and infection intensity were measured.
We expected prevalence and infection intensity to peak at 30°C in alignment with its peak growth in vitro (Palmer-Young et al., 2018. We also predicted that infection would be impeded by higher temperatures, due to a reduction in growth in vitro above 37°C (Palmer-Young et al., 2018 and an 81% decrease in infection intensity when bees were incubated at 37°C compared to 21°C (Palmer-Young et al., 2019). Finally, we predicted that prevalence and infection intensity would be higher after C. bombi was incubated for 10 compared to 60 min (Schmid-Hempel et al., 1999).
Crithidia bombi was obtained from two laboratory stock colonies of B. terrestris audax (Agralan). The parasite was originally acquired from post-hibernation spring queens of B. terrestris audax caught in Windsor Great Park (Surrey, UK) in March 2021, since when it has been continually cycled through laboratory colonies (Agralan).
Crithidia bombi has three lifecycle stages. One lifecycle stage is nonmotile (amastigote) and two lifecycle stages are motile (choanomastigote and promastigote; Figure 1). The prevalence of each lifecycle stage can change over the course of an infection (Logan et al., 2005).
Our pilot experiments also indicated that these temperatures and incubation durations were suitable (results in Appendix S1).

| Counting the number of healthy and unhealthy Crithidia bombi cells in the inoculum prior to inoculation
Pilot experiments (Appendix S1) identified C. bombi cells that appeared less healthy compared to others. Based on pilot observations, F I G U R E 1 Photo of study species: Buff-tailed bumblebee laboratory colony (Bombus terrestris audax, left) and Crithidia bombi at ×400 magnification under a phase contrast microscope (right). criteria for these cells were specified (Table 1, Figure 2). To assess whether the number of healthy and unhealthy C. bombi cells differed between treatments, the number of healthy and unhealthy cells in the inoculum were counted using an improved Neubauer haemocytometer at ×400 magnification using a phase contrast microscope.

| Inoculation
Individual workers were inoculated with a dose of 20,000 C. bombi cells. This dose was chosen to enable the assessment of prevalence and infection intensity, whilst maintaining field realism (Schmid-Hempel & Schmid-Hempel, 1993). Inoculation dose was also trialled in pilot experiments (see Appendix S1). On each day, two temperatures were tested, one for 10 min and one for 60 min. To prevent order effects, the order of temperatures was arranged such that a temperature was not tested twice on the same day and the order of temperatures was different for each incubation duration (Table S1).
On each day, 32 individuals (four per colony) were inoculated per temperature and duration of incubation combination. Therefore, 320 individuals were inoculated in total, 64 each day.
Individuals were removed from their colonies and weighed in preweighed vials to the nearest milligram (Scout SKX; Ohaus). Mass was used as a proxy for size since size can affect C. bombi infection intensity (Otterstatter & Thomson, 2006). Mass was used, rather than intertegular distance or wing marginal cell length, due to time constraints on inoculation days. We appreciate that body mass may be influenced by sugar consumption, but as all bees had equal exposure to ad libitum food prior to weighing, this seems unlikely to have a meaningful impact on results. Bees were housed individually in nicot cages (Becky's bees), which are cylindrical containers adapted from hair rollers to house bees (see Figure S1). Bees were starved for 2 h prior to inoculation. Faeces were collected from 20 individuals per C. bombi stock colony and purified using a modified triangulation protocol (Baron et al., 2014;Cole, 1970). The triangulation protocol involves systematically centrifuging the faecal samples to remove contaminants, such as pollen.  end of syringes were removed to allow access to the inoculum.
Individuals were left to drink for 4 h; if they had not consumed the entire droplet the individual was discarded from the experiment.

| Housing
Individuals were housed in nicot cages with sterile sugar solution (50% concentration) provided ad libitum via a 5-mL syringe attached to the base of the nicot cage (see Figure S1). As before, the end of syringes was removed to allow access to the sugar. Syringes were replaced every 3 days to prevent fungal growth.

| Measuring infection
One week after inoculation bees were put in specimen tubes and approximately 10 μL faecal samples were taken using microcapillary tubes. Faecal samples were viewed under a phase contrast microscope at ×400. Prevalence and infection intensity were measured using an improved Neubauer haemocytometer (see Appendix S1).
Prevalence was defined as the percentage of individuals infected

| Statistical analysis
Analyses were performed in RStudio "Prairie Trillium" (RStudio Team, 2022), R version 4.2.0 (R Core Team, 2022). All figures were created using the ggplot() function from the ggplot2 package (Wickham, 2016 (Arnqvist, 2020;Gelman & Hill, 2006). A likelihood ratio Chi-squared test and AIC values were used to compare reduced and full models. Model assumptions were checked graphically and using the DHARMa package (Hartig, 2022).
To investigate whether infection intensity was affected by parasite incubation temperature and duration a generalised linear model with a negative binomial error distribution was constructed using the glm.
nb function from the MASS package (Ripley et al., 2022). Only infected bees were included in this analysis. Infection intensity was measured as cells per microlitre rounded to the nearest integer. In the full model, temperature, incubation duration and their interaction were included as fixed effects, in addition to bee mass and colony, which were included as covariates. Model fit was assessed as described above.

| RE SULTS
A total of 289 bees were successfully inoculated and screened for infection. Some individuals were lost from the sample because they did not drink the entire inoculum (n = 12), died before screening (n = 11) or did not defecate during screening (n = 8).

| Do incubation temperature and duration affect the number of unhealthy cells in 1 μ L of inoculum?
Incubation temperature and duration did not affect the number

| Do incubation temperature and duration affect the prevalence of infection?
A model that removed colony and included temperature, incubation duration and their interaction, and bee mass had the best fit (see Appendix S1 for full model results). Prevalence of infection was not significantly affected by incubation temperature  Figure 4). Between 20 and 40°C, longer incubation slightly reduced prevalence, however, at 50°C, longer incubation drastically reduced prevalence to 0%. Body mass of individual bees did not significantly affect prevalence (b = 3.5, SE = 2.12, z = 1.65, p = .0993; Figures S8 and S9).

| Do incubation temperature and duration affect the intensity of infection in infected individuals?
A reduced model (including temperature, incubation duration as fixed effects, bee mass and colony as covariates) and excluding the interaction between temperature and incubation duration had the best fit. Temperature and incubation duration did not significantly affect infection intensity ( 2 4 = 311, p = .0785; 2 1 = 308, p = .0837; Figure 5). Figure 5 shows that there was little difference in infection intensity between treatment groups. Furthermore, bee body mass and colony did not affect infection intensity ( 2 1 = 306, p = .0921; χ 7 = 308, p = .0627).

F I G U R E 4
The prevalence of infection of Crithidia bombi in Bombus terrestris audax 1 week after infection. Prior to inoculation, C. bombi was exposed to five temperatures for two time periods. Pale green (left) indicates exposure to the temperature for 10 min and dark green (right) for 60 min. Error bars show 95% binomial confidence intervals. Sample sizes are given above each bar.

F I G U R E 3
The estimated percentage of the number of unhealthy Crithidia bombi cells in 1 μL of inoculum following incubation at five temperatures for 10 (pale green, left) and 60 min (dark green, right). Error bars show 95% binomial confidence intervals and sample sizes of the total number of counted cells are above each bar.

| Do incubation temperature and duration affect the presence of unhealthy Crithidia bombi cells in faeces?
The best model included temperature, the interaction between temperature and incubation duration, the quadratic term for temperature and its interaction, incubation duration and colony.  Figure 6). After incubation for 10 min, the prevalence of burst cells was more than two times higher at 10°C compared to 30°C, and three times higher at 50°C compared to 30°C (Figure 6). A similar pattern was seen after incubation for 60 min, with the lowest prevalence of burst cells observed at 30°C.

| DISCUSS ION
Overall, we found that below 50°C, C. bombi prevalence was not affected by incubation temperature or duration, indicating that the parasite remains infectious across a wide range of temperatures.
The interaction between incubation temperature and duration affected the prevalence of C. bombi at 50°C, resulting in no infections after 60 min. Furthermore, the intensity of infection was not affected by incubation duration or temperature. Despite being able to successfully infect and replicate once inside the host, unhealthy, burst C. bombi cells were more than twice as prevalent at 10°C compared to 30°C after incubation for 10 and 60 min, and more than three times as prevalent at 50°C compared to 30°C when incubated for 10 min. This suggests that temperature may cause physiological changes in the parasite, with the optimum temperature for C. bombi approximately 30°C. These changes appear to occur after infection because the incubation temperature and duration did not affect the presence of unhealthy cells in the inoculum.
The effect of incubation duration on prevalence of infection indicates that the duration of time C. bombi spends outside the host affects its ability to infect the next host. Between 20 and 40°C, incubating the inoculum for 60 min resulted in a ~4% decrease in prevalence at a given temperature, compared to incubating the inoculum for 10 min. This is a relatively small decrease when prevalence in worker populations in the field is high, for example 77.7% (Popp et al., 2012). However, towards the end of the summer prevalence tends to fall, for example, to 3.9% in August (Parsche & Lattorff, 2018), and at this time a 4% decrease would have a much larger effect on overall prevalence. In contrast to our results, Schmid- Furthermore, when considering the duration of time that C. bombi is exposed to the environment, the rate of visitation by bumblebee foragers needs to be considered. The visitation rate of different plants varies with multiple floral design features, such as flower colour F I G U R E 6 The prevalence of infected Bombus terrestris audax with unhealthy Crithidia bombi cells (based on criteria in Table 1 and Figure 2) 1 week after being infected with C. bombi. Prior to inoculation, C. bombi was exposed to five temperatures for 10 (pale green bars, left) or 60 min (dark green bars, right). Error bars show 95% binomial confidence intervals. Sample sizes are above each bar. (Stanton, 1987) and the rate of nectar production (Mitchell, 1994).
Therefore, plants with shorter visitation rates may be more likely to transmit C. bombi, particularly, for C. bombi cells shed on the surface of floral structures rather than in the nectar.
The response of C. bombi prevalence to temperature exposure appears not to be linear but a threshold response. The threshold for infectivity was surpassed at 50°C after 60 min, as no bees were successfully infected after this treatment. Lack of infectivity at higher temperatures could be due to the denaturation of enzymes and proteins (Copeland, 2000). The fact that infection intensity and the presence of the three lifecycle stages were not significantly affected by temperature shows that once established in a host C. bombi was able to successfully replicate, irrespective of its previous exposure. Temperature can affect the separate stages of parasite transmission to a host differently. Increased temperatures, for example, have been shown to increase the ability of the trematode, Ribeiroia ondatrae, to penetrate host skin, but reduces the ability of trematode larvae, called cercariae, to encyst after skin penetration (Paull et al., 2012). Furthermore, the increase in the proportion of unhealthy C. bombi cells at 50°C compared to 30°C indicates that morphological changes occurred following temperature exposure, but these changes were not visible immediately, because the presence of unhealthy cells in the inoculum did not vary between treatments. When exposed to higher temperatures another trypanosome, Leishmania spp, exhibits morphological changes to the cell structure and the parasite loses its ability to multiply (Zilberstein & Shapira, 1994). In Leishmania spp. these changes are reversible, and the parasite regains its ability to multiply when the temperature falls. This appears not to be true for C. bombi as exposure to 50°C was temporary. The lack of infectivity after prolonged exposure to 50°C suggests that C. bombi transmission may be curtailed in climates and seasons with very high temperatures.
Furthermore, due to climate change, temperatures above 40°C are becoming increasingly common and more prolonged (Coumou et al., 2013). For example, in the summer of 2022 temperatures surpassed 40°C in the United Kingdom (Kendon, 2022). Consequently, C. bombi is expected to be exposed to temperatures above 40°C more frequently than previously. infection intensity of C. bombi declined by 81% when hosts were incubated at 37°C compared to 21°C. These studies were conducted in the lab, in a non-stressful environment with food ad libitum, whilst, in the field, food may be less available. Nutritionally stressed bees exhibit higher mortality when infected with C. bombi (Brown et al., 2000), and therefore, the effects of temperature on host susceptibility may be different in the field. However, a more important consideration for understanding whether the response of the parasite to high temperatures is relevant for transmission is host behaviour. At temperatures above 24°C, bumblebee flight ability decreases (Kenna et al., 2021) and at extremely high temperatures, such as above 40°C, bumblebees may reach their thermal maximum and enter heat stupor (Martinet et al., 2021). If bumblebees are unable to fly at temperatures above 40°C, transmission will not be affected by reduced C. bombi infectivity because the bumblebees will not encounter the parasite. However, thermal tolerance varies between species, with some Mediterranean sub-species, such as B. xanthopus, able to withstand extended periods of time at these temperatures (Martinet et al., 2021;Oyen et al., 2016). It is unclear whether these species can fly over 40°C or whether they are just more resistant to heat stupor (Martinet et al., 2021). If they can fly, extreme temperatures may reduce pathogen transmission to heattolerant species.
In contrast, at lower temperatures, C. bombi infection ability was not impeded, despite a higher proportion of cells looking unhealthy at 10°C compared to 30°C. The fact that C. bombi infectivity was approximately constant between 10 and 40°C suggests that temperature does not play a major role in the seasonal dynamics of C. bombi prevalence, which peaks in early summer in the northern hemisphere (Parsche & Lattorff, 2018;Popp et al., 2012). Based on our results C. bombi is likely equally as infective to emerging spring queens at 10°C as it is to foraging workers in the peak of summer.
Rather, population demographic changes as the summer progresses likely play a major role in seasonal peaks in prevalence (Parsche & Lattorff, 2018;Popp et al., 2012). However, seasonal fluctuations in prevalence have only been studied in limited parts of C. bombi's range, so it would be interesting to test whether seasonal patterns in prevalence are the same across hotter regions, where temperatures surpass 40°C.
Finally, here we only looked at the effect of temperature on infectivity, whereas, when outside the host, C. bombi may be exposed to varying levels of humidity, UV radiation and potentially precipitation. These factors may also affect infectivity in isolation or may interact to alter infectivity. Climatic factors other than temperature, have been shown to affect the prevalence of some bumblebee pathogens including Vairmorpha bombi (Manlik et al., 2023;McNeil et al., 2020), C. bombi and C. expoeki (Ivers et al., 2022). Specifically, the prevalence of V. bombi can be affected by temperature, humidity, precipitation and cloud cover (Manlik et al., 2023). In addition, some have found a positive correlation between V. bombi prevalence and spring precipitation (McNeil et al., 2020). It is likely that UV radiation affects C. bombi survival because C. bombi survives for shorter time periods on sunny compared to shaded flowers (Figueroa et al., 2019). Sunlight-level UV radiation has been shown to increase the mortality of nematode larvae (van Dijk et al., 2009), emphasising the importance of considering all aspects of climate variability. Exposure to different environmental variables whilst C. bombi is on flowers compared to when on nest material may mean C. bombi is more transmissible in one environment compared to the other. For example, if UV radiation reduces C. bombi survival, transmission on flowers may be lower than transmission from contaminated nest material within the colony at the same temperature.

| CON CLUS IONS
In conclusion, our study showed that the parasite, C. bombi, was less infective the longer it spent outside the host under experimental conditions. This may suggest that the likelihood of transmission in the wild is higher on flowers with shorter visitation times. In addition, the infectivity of the parasite was not affected by exposure to temperatures between 10 and 40°C. Crithidia bombi was no longer infective after exposure to 50°C for 60 min.
However, it is unlikely that many bumblebee species would fly and, therefore, encounter the parasite at this temperature. Infection intensity was not affected by temperature, however, extreme temperatures appear to induce morphological changes to C. bombi cells in the faeces of infected individuals. We investigated the effect of one climate variable on parasite infectivity when in reality multiple climate factors will vary in the environment. Assessing the effect of these on hosts and parasites in isolation, and when interacting, will further our understanding of the epidemiology of host-parasite interactions across different climates and help to predict the effects of climate change in the future.

ACK N O WLE D G E M ENTS
None.

This work was funded by a Royal Holloway University of London
Scholarship.

CO N FLI C T O F I NTER E S T S TATEM ENT
We declare no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
DOI for data: http://doi.org/10.5281/zenodo.7760010.