Temperature and host diet jointly influence the outcome of infection in a Daphnia- fungal parasite system

1. Climate change has the potential to shape the future of infectious diseases, both directly and indirectly. In aquatic systems, for example, elevated temperatures can modulate the infectivity of waterborne parasites and affect the immune response of zooplanktonic hosts. Moreover, lake warming causes shifts in the communities of primary producers towards cyanobacterial dominance, thus lowering the quality of zooplankton diet. This may further affect host fitness, resulting in suboptimal resources available for parasite growth.

. Elevated temperatures have been shown to modulate the intensity, timing, and transmission of infectious diseases, including bacterial pathogens of nematodes (Stirling, 1981), microsporidia infecting honeybees (Martín-Hernández et al., 2009) and human malaria (Paaijmans et al., 2010). Further signs of a rapidly changing world are shifts in food-web structures: climate change may alter the community composition of key trophic groups (Petchey, McPhearson, Casey, & Morin, 1999), induce trophic mismatch through changes in phenology (Edwards & Richardson, 2004), alter the primary productivity, and even reduce the total biomass in a food web (O'Connor, Piehler, Leech, Anton, & Bruno, 2009).
However, in addition to temperature, diet quality and nutrient uptake are equally important drivers of the metabolic processes governing host immunity (Landolt, 1989;Las Heras et al., 2019) and parasite performance (Arostegui, Hovel, & Quinn, 2018;Crompton, 1987). For instance, protein supplementation of ruminants has been shown to increase resistance to gastrointestinal nematodes (Coop & Holmes, 1996), while experimental increases in nutrient concentrations enhanced the severity of different coral pathogens (Bruno, Petes, Drew Harvell, & Hettinger, 2003). Although hosts may benefit from high-quality diets in the form of enhanced immune response or increased fecundity, higher host densities may in turn provide parasites with a larger pool of potential hosts, each serving as a nutritive resource to their own parasite (Pike, Lythgoe, & King, 2019). Thus, in the context of epidemiological studies, temperature increase and food-web alterations are two aspects of climate change that should be regarded as related phenomena.
Considering the central role of Daphnia in the trophic structure of aquatic food webs, any factor modulating the abundance and composition of zooplankton populations might lead to detrimental effects on the functioning of freshwater ecosystems. In addition to the negative effects associated with the dominance of cyanobacteria, a wide range of microparasites such as microsporidia, fungi and bacteria represent another threat to Daphnia hosts (Ebert, 2005).
Most of these parasites negatively affect Daphnia survival and reproduction (Ebert, 2005;Green, 1974) and can reduce their abundance to such levels that control of phytoplankton by grazing is inhibited (Duffy, 2007). Predicting the overall direction of Daphnia parasitism under a climate change scenario is challenging, as warming may trigger cascading effects that modulate disease outcomes in complex and intricate ways. First, rising temperatures could directly alter zooplankton susceptibility to infection (Mitchell, Rogers, Little, & Read, 2005;Schoebel, Tellenbach, Spaak, & Wolinska, 2011), as well as the physiology of their parasites (Shocket et al., 2018;Vale, Stjernman, & Little, 2008). Second, the resulting proliferation and dominance of cyanobacteria might weaken host defences due to reduced nutrient uptake or cyanotoxin-induced stress. This has been suggested for Daphnia populations infected by the gut parasite Caullerya mesnili, as cyanobacterial density positively correlated with the occurrence of epidemics (Tellenbach et al., 2016). However, by producing antibiotic or antifungal effects, cyanobacteria may also interfere with pathogens (Abed, Dobretsov & Sudesh, 2009;Singh, Tiwari, Rai, & Mohapatra, 2011). Such medicinal properties have been suggested for the common cyanobacterium Microcystis aeruginosa against two parasites of Daphnia: the viral agent of white fat cell disease (Coopman, Muylaert, Lange, Reyserhove, & Decaestecker, 2014) and the yeast Metschnikowia bicuspidata (Sánchez, Huntley, Duffy, & Hunter, 2019). Overall, despite the substantial effort to relate the fitness of Daphnia parasites to single factors, such as food quality (Hall, Knight, et al., 2009a;Sánchez et al., 2019), nutrient availability (Frost, Ebert, & Smith, 2008;Narr, Ebert, Bastille-Rousseau, & Frost, 2019) and water temperature (Cuco, Castro, Gonçalves, Wolinska, & Abrantes, 2018;Vale et al., 2008), the combined effects of these stressors remain relatively unexplored in this system (but see Garbutt, Scholefield, Vale, & Little, 2014). As cyanobacteria blooms and heat waves are concurrent phenomena in nature (Joehnk et al., 2008), a comprehensive approach is required to make better epidemiological predictions in freshwater ecosystems.
To explore how elevated water temperature and decreased food quality interact at the host-parasite interface, we used two Daphnia genotypes in a fully factorial design including three food sources of varying quality (Scenedesmus obliquus as high-quality green algae/M. aeruginosa or Planktothrix agardhii as morphologically distinct, low-quality cyanobacteria), two levels of temperature (standard/elevated) and infection by the parasitic yeast M. bicuspidata (control/exposed). We recorded the proportion of successful infections following exposure (parasite infectivity) and the number of spores produced at host death (parasite reproduction). We combined those metrics into an estimate of parasite fitness (net parasite output, which conveys the expected number of transmission stages contributing to the next generation of parasites). Fitness parameters (average lifespan, fecundity, and body size) were measured to quantify the effects of environmental conditions and infection on Daphnia hosts. We predicted a generally enhancing effect of elevated temperature, but a detrimental effect of low food quality on net parasite output, which might result in a potential equilibrium when both stressors are combined.

| Study system
The zooplankter Daphnia (Crustacea: Cladocera) was used as the focal host. Daphnia reproduce through cyclical parthenogenesis, allowing for the inclusion of distinct clonal lines in the experimental design (Ebert, 2005). Two genotypes of Daphnia longispina × galeata hybrids (AMME_12 and AMME_51) were selected randomly from a wider collection of clonal lines isolated from Ammersee, Germany. Hybrids belonging to the D. longispina species complex are common and sometimes dominant inhabitants of permanent water bodies across the world (Griebel et al., 2015;Keller, Wolinska, Manca, & Spaak, 2008), being also able to colonise intermediate habitats that are not shared by their respective progenitor species (Ma, Hu, Smilauer, Yin, & Wolinska, 2018).
Infection takes place upon ingestion of spores by water-filtering hosts. Mature, needle-shaped spores pierce the gut wall before reaching the haemolymph (Codreanu & Codreanu-Balcescu, 1981).
Infection symptoms become clearly visible after 9-10 days, when the host's body cavity starts to fill with the ascus stage (Stewart Merrill & Cáceres, 2018). Spore release occurs after host death, once the cuticle starts to decompose, allowing for parasite spores to be ingested by new hosts. A single M. bicuspidata strain was used, also isolated from Ammersee. This strain was later propagated on a laboratory-reared Daphnia magna clone (Hesse, Engelbrecht, Laforsch, & Wolinska, 2012). Due to its low host specificity, the parasite can be raised on D. magna-a larger host species which conveniently provides high spore output upon death-and later used to infect other Daphnia species (Cuco et al., 2018;Hesse et al., 2012).
Three phytoplankton species were used as different food sources for the host: the unicellular green alga S. obliquus (long-standing laboratory culture used as standard food for Daphnia), the coccoid cyanobacterium M. aeruginosa (MaGr01, isolated from Greifensee in Switzerland; Tellenbach et al., 2016) and the filamentous cyanobacterium P. agardhii (NIVA-CYA 630, isolated from Lake Lyseren in Norway; https ://niva-cca.no). Both cyanobacteria species were selected as common bloom-forming taxa (Reynolds & Wakby, 1975;WHO 2009). Laboratory cultures of MaGr01 lost their colonial morphology, and single cells display an optimal size range for Daphnia ingestion. While both MaGr01 and NIVA-CYA 630 have been confirmed to produce microcystin (Rohrlack et al., 2008;Tellenbach et al., 2016), Planktothrix also displays a filamentous morphology, which reduces its susceptibility to grazing (Gliwicz, 1977;Lampert, 1987

| Experimental setup
Prior to the start of the experiment, the two D. longispina × galeata two temperatures (standard/elevated) and two infection treatments (control/exposed to Metschnikowia). Ten replicates were set up for unexposed Daphnia and 20 replicates for exposed ones, accounting for a total of 360 experimental units. To establish similar exposure conditions across temperature treatments, a unit of physiological time was employed, namely degree-days (calculated as the product of real-time in days and temperature in °C). This was used to account for relatively faster growth at 23°C, which leads to higher filtration rate due to larger body sizes, and thus higher spore uptake (Burns, 1969;Hall et al., 2007).
Experimental Daphnia were born within a 48-hr time span, after which mothers were removed from the common jars. At degree-day 95 (day 5 at 19°C/day 4 at 23°C), experimental subjects were transferred to individual jars containing 5 ml of fresh culture medium. At degree-day 115 (day 6 at 19°C/day 5 at 23°C), all jars were checked for early mortality and Daphnia were replaced if needed.
Experimental jars were then inoculated with a suspension obtained by crushing the same amount of tissue from either infected or uninfected D. magna in the exposed and control solutions, respectively.
To maximise infection success, this exposure protocol was repeated after 2 days, as in Yin, Laforsch, Lohr, and Wolinska (2011) (applied concentrations: 700 spores/ml and 550 spores/ml for the first and second exposure events, respectively). To determine spore concentrations, the homogenised suspension was loaded twice (2 × 10 µl) on an Improved Neubauer counting chamber. Total spore yield was estimated from the mean number of mature spores counted in four squares of 1 µl capacity, across two independent loads.
During the first few days following the onset of the experiment, Daphnia were fed with 1 mg C/L of S. obliquus. Food quantity was reduced to 0.5 mg C/L once Daphnia were transferred into individual jars. To maximise infection success, animals were not fed during the first day of exposure (low food density was shown to promote spore uptake, Hall et al., 2007). Daphnia were separated into their respective food treatments the day following the first exposure event, i.e. at degree-day 135 (day 7 at 19°C/day 6 at 23°C). In the Scenedesmus food treatment, animals were fed daily with 0.5 mg C/L of S. obliquus. In the other two treatments, a food mixture was used in which either Microcystis or Planktothrix contributed 75% of the total amount of carbon, with Scenedesmus contributing the remaining 25%. The correlation between optical density and carbon content for each phytoplankton taxon was established and used to prepare food suspensions accordingly. Following the second exposure event, the experimental volume was raised to 15 ml (day 9 at 19°C/day 8 at 23°C). From this point onward, individuals were transferred to fresh medium every 4 days. Neonates were counted and removed daily, with those from the second clutch kept frozen for body size determination (−20°C). All exposed individuals which died after 8 days post-exposure (earliest observation of infection) were fixed in 3% formaldehyde. As no further deaths were observed after 27 days into the experiment, it was terminated soon after. All surviving individuals were fixed in 3% formaldehyde.

| Parasite fitness
Parasite infectivity (calculated as the proportion of successfully infected individuals) was assessed by checking fixed animals for the presence of parasite spores under a dissecting microscope (30× magnification). Parasite reproduction (the number of spores produced until host death, calculated individually per infected host) was estimated from a suspension of crushed infected Daphnia using a counting chamber (see Experimental setup). Conveniently, parasite reproduction was shown to be a good estimate of transmission rates in Daphnia (Izhar & Ben-Ami, 2015). To combine these intermediate fitness components into a single metric that encompasses parasite success, we devised the net parasite output. For the parasite to contribute to the next generation, two conditions need to be met. First, the host has to survive long enough for the parasite to complete its infection cycle (defined here as host survival probability). Second, the surviving host has to become terminally infected (this probability was conveyed as parasite infectivity). Consequently, net parasite output is defined as the product of host survival probability, parasite infectivity and parasite reproduction. Host survival and parasite infectivity were computed for each combination of food quality, temperature and host genotype (12 treatments), out of 20 Daphnia which were exposed to the parasite in each treatment (Table S1).

| Host fitness
Age at death was recorded for each individual Daphnia that died starting from day 7 at 19°C and day 6 at 23°C (after the initial replacement of early deaths due to background mortality). Animals that were fixed in formaldehyde on the last experimental day were considered to have died at that time (none of these individuals were found to be infected). Body size was recorded for juveniles from the second clutch and for adult Daphnia which were retrieved on the last experimental day, including those that were exposed but not infected (see Figure S1) and those from the control treatment (due to age differences, body size was otherwise not recorded for animals that died from infection). Daphnia were measured under a dissecting microscope using with r as the rate of population increase (/day), x the age class in days, l x the probability of surviving to age x, and m x the fecundity at age x (Cuco et al., 2018;McCallum, 2000). Pseudovalues were generated by jackknifing and reassigned as individual values for each replicate in a given treatment (Meyer, Ingersoll, McDonald, & Boyce, 1986). Prior to inspection of infection status, spore yield, body size of adults and juveniles, all samples were assigned random numbers and relabelled to ensure blind assessment.

| Data analysis
Data were analysed using R version 3.6.0 (R Core Team, 2019).
Graphical outputs were produced using the ggplot2 (Wickham, 2016) and Hmisc (Harrell & Harrell, 2019) packages. Analysis of variance (F-test or χ 2 test) was performed with the car package (Fox et al., 2012) using type III sums-of-squares. Whenever no significant interaction was recorded or missing values led to aliased coefficients in a model, type II sums-of-squares were used instead. Model selection was then performed by a stepwise regression approach based on Akaike information criterion.

| Host fitness
Age at death, fecundity (the total number of offspring) and growth rate (the per capita intrinsic rate of increase, r) were analysed using generalised linear models with Food, Temperature, Infection, and Clone as explanatory variables, assuming a negative binomial distribution (package MASS, function glm.nb) or γ-distribution of the residuals. Body size of adults and body size of juveniles (averaged per mother) were analysed using linear models with Food, Temperature, Infection, and Clone as explanatory variables. Preliminary analyses were run with all four factors (Food, Temperature, Infection, and Clone) as main effects only. If no significant effect of Clone was detected, this factor was deleted from the subsequent analysis and a three-way ANOVA was performed instead, including all interactions between the remaining factors. Since exposed Daphnia could only be confirmed as infected after surviving at least 8 days after exposure, early deaths were pooled together with terminally infected individuals in order to be compared with the control treatment ( Figure S1).

| Parasite fitness
Out of 240 Daphnia exposed to Metschnikowia spores, seven individuals were lost due to handling error and 78 individuals died before day 8 post-exposure (categorised as early death, see Figure   S1). Among the 155 remaining individuals, 98 were confirmed as infected and 57 remained uninfected (categorised as infected and  clone AMME_12 only (significant Temperature × Clone interaction). Parasite reproduction within infected hosts was highest in the Scenedesmus treatment (significant Food effect, Figure 1c).
However, this effect of host diet was clone dependent. For instance, under a Microcystis diet, parasite reproduction was higher on clone AMME_51 (significant Food × Clone interaction). Net parasite output was generally higher when the host was maintained on the high-quality diet, Scenedesmus (significant Food effect, Figure 1d). However, when clone AMME_51 was exposed to elevated temperature under a Scenedesmus diet, net parasite output was greatly reduced, being surpassed by the low-quality Microcystis diet (significant Food × Temperature × Clone interaction). Moreover, parasite output on clone AMME_51 was higher than on clone AMME_12 under a Microcystis diet (significant Food × Clone interaction).

| Host fitness
Preliminary analyses revealed no significant effect of Daphnia genotype on any of the variables related to host fitness. Consequently, this factor was removed from the analyses. Host lifespan was greatly reduced by infection (significant Infection effect,  Figure S2a). However, exposed hosts maintained on a Scenedesmus diet, which did not become infected (exposed but not infected) reached smaller adult sizes than their control conspecifics (significant Food × Infection interaction). As opposed to adult Daphnia, the body size of juveniles from the second clutch was highest under a Microcystis diet (significant Food effect, Figure S2b). Neither temperature nor infection influenced the size of offspring (Table S2).

F I G U R E 1
Comparison of traits relating to infection success of the yeast parasite, Metschnikowia bicuspidata. Two Daphnia genotypes (AMME_12, AMME_51) were exposed to the parasite under two temperatures (19°C, 23°C) and three food treatments (Scenedesmus, Microcystis, Planktothrix). (a) Host survival (proportion of hosts which survived until day 8 post-exposure); (b) parasite infectivity (proportion of successful infections); (c) parasite reproduction (number of spores produced); (d) net parasite output (product of the previous three variables). Error bars represent the standard error of the mean. Due to high mortality of AMME_51 in the Planktothrix × 23°C treatment, parasite reproduction could not be estimated for this combination: only one individual survived until parasite inspection, but was not infected (Table S1)

By exposing Daphnia hosts to the common waterborne parasite
Metschnikowia, our aim was to gain insight into how specific combinations of temperature and diets (representing future environmental disturbances in warmed lakes) may affect key traits of this host-parasite system. To enable ecologically relevant predictions regarding the potential for disease spread in future environments, we chose to focus on two synthetic variables: the net parasite output per exposed host, as well as the population growth rate of the host (r), which ensures the renewal of new hosts for the parasite to infect.

| Parasite fitness
Food quality appeared to be the main driver of net parasite output, TA B L E 2 Three-way ANOVA testing for fixed effects of food quality, temperature, infection, and their interactions on life history parameters of the host (two Daphnia genotypes). The dataset entries describe which subsets of data were compared as levels of the Infection factor (see also Figure S1).  (Crompton, 1987;Hall, Simonis, Nisbet, Tessier, & Cáceres, 2009b). Similarly, spore production of Metschnikowia was hampered when its host was fed with field-collected, poor-quality algae as opposed to Ankistrodesmus falcatus (Hall, Knight, et al., 2009a), and was also found to be lower in lakes with high C:P ratios (Civitello et al., 2015). In addition to food quality, restricted quantities of a standard resource were also found to reduce growth of another Daphnia parasite, the bacterium Pasteuria ramosa (Frost et al., 2008;Stjernman & Little, 2011). Arguably, rather than a consequence of low food quality per se (i.e. lack of sterols and long-chain poly-unsaturated fatty acids in cyanobacteria; Gerphagnon et al., 2018), our results could also be partially explained by the reported antifungal properties of M. aeruginosa (Sánchez et al., 2019). While the genus Planktothrix has not been tested for its antifungal properties against Daphnia parasites, it produces a wide array of bioactive secondary metabolites (Kurmayer, Deng, & Entfellner, 2016), that are likely to be involved in the defence against fungal chytrid parasites (Rohrlack, Christiansen, & Kurmayer, 2013;Sønstebø & Rohrlack, 2011).
While host diet turned out to be a preponderant driver of parasite fitness, the effects of temperature were less straightforward, manifesting mostly as complex interactions with host genotype or food quality, rather than as main effects. The absence of a general effect of temperature was surprising, as elevated temperatures are associated with an increase in metabolic rates (O'Connor & Bernhardt, 2018). Thus, high temperatures may increase the filtration rate of zooplankton, thereby facilitating the uptake of fungal spores (Shocket et al., 2018). Based on such findings, we expected both Daphnia genotypes to display increased susceptibility to the fungal parasite at 23°C. Instead, one host genotype became more easily infected when exposed to elevated temperature, while the other experienced compromised survival and reduced spore yield, leading to inferior parasite success. Such host genotype-specific responses to elevated temperature have also been discovered for other pathogens of Daphnia (Garbutt et al., 2014;Schoebel et al., 2011), as well as across many other host-parasite systems (reviewed in Wolinska & King, 2009

| Host fitness
In the absence of the parasite, cyanobacterial diets severely reduced host growth, due to a combination of impaired offspring production (both cyanobacterial species) and compromised survival (Planktothrix). The combination of high levels of host fecundity and efficient net parasite output under a Scenedesmus diet suggest that high-quality, green algal diets are more likely to promote epidemic outbreaks than the typical cyanobacteria occurring under bloom conditions. Furthermore, hosts exposed to a combination of elevated temperatures and high densities of toxic cyanobacteria, such as Planktothrix, might not live long enough to ensure transmission of the parasite. If such conditions became more prevalent as a result of climate change (Paerl & Paul, 2012;Paul, 2008), selection for faster replicating parasite strains might arguably occur in the wild. While the fungal parasite used in this experiment displays limited genetic diversity in natural populations Searle et al., 2015;Wolinska, Giessler, & Koerner, 2009) and did not respond to a selection experiment (Auld, Hall, Housley Ochs, Sebastian, & Duffy, 2014), such evolutionary responses could still apply to other parasites of Daphnia with higher evolutionary potentials, such as the bacterium P. ramosa (Ebert et al., 2016).

| CON CLUS ION
By investigating the main and interactive effects of temperature and host diet in the Daphnia-Metschnikowia system, we conclude that elevated temperature does not universally enhance parasite fitness.
Instead, climate change is expected to promote the dominance of poor-quality algae and favour conditions of suboptimal nutrition in zooplanktonic hosts. This implies a reduction of exploitable resources for the parasite, resulting in decreased output, as an indirect effect of climate change. Distinct food sources appear to modulate host and parasite fitness in diverging ways, depending on host genotype and temperature. Such discrepancies suggest that toxic blooms might have different consequences depending on which cyanobacterial taxa become dominant when outbreaks occur. However, none of the tested cyanobacteria seem to enhance parasite epidemics, as they reduced host growth rates to negligible levels. Given the complex interactions that can arise between specific host diets and temperature conditions, the inclusion of both environmental factors in future experimental or modelling work on zooplankton pathologies seems pertinent and necessary.

ACK N OWLED G EM ENTS
This work was supported by a joint German-Israeli project (WO 1587/8-1 to J.W., 0604317501 to F.B.A.) and one other project (WO 1587/6-1 to J.W.), both funded by the German Science Foundation.
We would like to thank Ursula Newen for the maintenance of

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.