Temperature modifies trait-mediated infection outcomes in a Daphnia–fungal parasite system

Abstract

One major concern related to climate change is that elevated temperatures will drive increases in parasite outbreaks. Increasing temperature is known to alter host traits and host–parasite interactions, but we know relatively little about how these are connected mechanistically—that is, about how warmer temperatures impact the relationship between epidemiologically relevant host traits and infection outcomes. Here, we used a zooplankton–fungus (Daphnia dentifera–Metschnikowia bicuspidata) disease system to experimentally investigate how temperature impacted physical barriers to infection and cellular immune responses. We found that Daphnia reared at warmer temperatures had more robust physical barriers to infection but decreased cellular immune responses during the initial infection process. Infected hosts at warmer temperatures also suffered greater reductions in fecundity and lifespan. Furthermore, the relationship between a key trait—gut epithelium thickness, a physical barrier—and the likelihood of terminal infection reversed at warmer temperatures. Together, our results highlight the complex ways that temperatures can modulate host–parasite interactions and show that different defense components can have qualitatively different responses to warmer temperatures, highlighting the importance of considering key host traits when predicting disease dynamics in a warmer world.

This article is part of the theme issue ‘Infectious disease ecology and evolution in a changing world’.

1. Introduction

Elevated temperature can have major impacts on host–parasite interactions and infection outcomes, shaping disease epidemics [1]. A growing body of work demonstrates that parasites often outperform their hosts at elevated temperatures, suggesting that a warmer world will be sicker [2–4]. However, this outcome is not universal (e.g. [1,2]), and, at present, it is challenging to predict the impact of warmer temperatures on a particular host–parasite interaction. Changing thermal regimes, such as increases in mean temperatures and extreme temperature events, can alter morphological, physiological and chemical traits of individuals [5–7]. However, most studies have focused on the effect of climate warming on parasite development and disease transmission [7,8]; empirical tests of how temperatures mediate individual host immune parameters that influence infection outcomes are still in their infancy [9,10].

Host traits, including physical barriers and chemical and immunological responses, play crucial roles in determining infection outcomes [11]. Recent work suggests that decomposing different functional steps of host–parasite interactions can allow for a better understanding of infection outcomes. Specifically, this will help reveal the mechanisms of how different forms of host defense mediate host susceptibility and resistance [12,13]. Susceptibility to infection is known to vary with temperature due to changes in individual physiological and immune responses [14,15], such that changing temperatures can have major fitness consequences for both hosts and parasites [16,17]. Furthermore, thermal stress can induce trade-offs between defenses against parasites and life-history traits, such as between reproduction and infection resistance [18]. However, despite the importance of understanding how changing environments influence host susceptibility to parasites, we are still uncertain about how the traits determining host susceptibility are impacted by warmer temperatures.

The crustacean zooplankton grazers Daphnia sp. play a significant role in freshwater food webs, where they consume algae and are in turn consumed by larger predators [19]. Daphnia are also frequently infected by several microbial parasite species [20], which can impact Daphnia population dynamics and have cascading impacts on food webs [21]. The yeast Metschnikowia bicuspidata (Ascomycota: Saccharomycetales) is a generalist parasite that is commonly found infecting many Daphnia species (as well as other Cladocerans) in nature [20,22–25]. In stratified lakes in the Midwestern United States, infection by M. bicuspidata typically happens annually during late summer/early autumn and can reach up to 60% prevalence in some populations [26,27]. The infection process begins when the parasite spores are ingested by the hosts during filter feeding, but not all hosts exposed to parasites become infected successfully. Once consumed, successful infection requires that the needle-shaped spores pierce the gut epithelium and enter the host body cavity (figure 1). Once in the body cavity, the parasite can begin to develop and reproduce. Some hosts recover after this early stage of infection, but others develop a ‘terminal infection’ (sensu [12,28]) where the host body is filled with conidia and/or asci, which leads to host death; once hosts reach this stage, recovery is no longer possible [12,28]. The next generation of spores are released from the dead hosts into the environment, where they are consumed by new hosts, completing the infection cycle [20].

Figure 1.

Infection process of the fungal parasite Metschnikowia bicuspidata within the host Daphnia dentifera. (a) A female host exposed to the parasite for 24 h. (b) M. bicuspidata are frequently found penetrating into the body cavity through the anterior and posterior gut epithelium. The image in (b) corresponds to the region surrounded by the red box in (a). The right and left arrows in (b) indicate haemocoel spores with and without haemocyte attachment, respectively. Photo credit: Syuan-Jyun Sun. (Online version in colour.)

Thus, in this system, host susceptibility to parasite infection is governed by two fundamental host traits: (1) a physical barrier and (2) cellular immune defenses [12]. First, the gut epithelium functions as a barrier to spores, with epithelial thickness influencing whether spores penetrate into the body cavity [12]. Such physical barriers to infection are common. In Daphnia, a key step in infections by the castrating bacterium Pasteuria ramosa is the attachment of the spore to the host oesophagus, after which it penetrates into the host body [29,30]. More generally, the peritrophic membrane serves as a barrier to infection in a wide variety of animals, including insects [31], and physical barriers to infection are common across taxa (e.g. plant cuticles [32] and skin in amphibians [33]). While physical barriers are a common form of defense, the impact of elevated temperature on this form of defense has largely been ignored. Second, spores that successfully penetrate the gut epithelium (referred to as ‘haemocoel spores’ because they enter the host haemocoel after penetrating the gut) face host internal defenses, with haemocytes (immune cells) upregulated and recruited to the site of infection ([9,25], figure 1b). Prior work has explored how immune responses are impacted by elevated temperatures; for example, fruit flies and butterflies both had reduced immune function at warmer temperatures [34,35]. Here, we consider how temperature impacts not only cellular immune defenses but also barrier defenses.

In this study, we examine the effect of warmer temperatures on the infection process and the consequences for terminal infection outcomes. Specifically, we used the host D. dentifera and its fungal parasite M. bicuspidata as a host–parasite model system to experimentally test how warmer temperatures alter key host traits, and, in turn, how this mediates infection outcomes. Daphnia spp. are a well-established ecological model for climate change and disease ecology studies [19,36,37]. We began by exposing individual D. dentifera to M. bicuspidata spores in laboratory conditions at control temperatures (20°C) and warmer temperatures (24°C), the former being selected to match the natural epilimnetic temperature range when the epidemics start [38]. During the infection stage, we investigated how temperature modified key host traits including gut epithelium thickness and haemocyte responses to attacking spores. We also evaluated the fitness consequences of temperature and terminal infection (defined as the presence of asci in the host haemolymph; [12]) on both host and parasite. We hypothesized that gut epithelium thickness and cellular immune responses determine whether hosts ultimately develop terminal infections, and that temperature will influence these traits and, therefore, the likelihood of terminal infection. More specifically, we predicted that warmer temperatures would hinder physical barrier and/or cellular immune defense against parasites, thus resulting in increased terminal infection and reduced host fitness.

2. Methods

(a) . Study species: origin and maintenance

Daphnia dentifera is a cyclically parthenogenetic crustacean commonly found in freshwater lakes in temperate North America [27]. Metschnikowia bicuspidata is an ascomycete yeast that commonly causes epidemics in D. dentifera [27]. We used the ‘Standard’ genotype of D. dentifera (originally isolated from a lake in Barry County, Michigan) and the ‘Standard’ isolate of the fungal parasite M. bicuspidata (Baker Lake in Barry County, Michigan). The D. dentifera stock was maintained as small populations in 150 ml beakers (5 animals per beaker, 40 beakers in total) filled with 100 ml filtered lake water. All animals were fed three times a week with phytoplankton food (Ankistrodesmus falcatus, 20 000 cells ml⁻¹). Prior to the experiments, Daphnia were reared under standardized laboratory conditions (on a 16:8 photoperiod at 22°C) for three generations. The M. bicuspidata culture was maintained by infecting 75 D. dentifera (two weeks to a month old) with M. bicuspidata (using spores obtained by grinding up infected D. dentifera that had been collected 1–2 weeks prior and stored in a refrigerator before use) at a density of 250 spores ml⁻¹ in a 1000 ml Erlenmeyer flask filled with 900 ml filtered lake water. The culture was fed three times a week (A. falcatus, 20 000 cells ml⁻¹) and checked every 10 days for infected D. dentifera, which were then placed in a 1.5 ml tube of 100 µl filtered lake water and stored in a refrigerator before use.

(b) . Experimental design

To test the effect of increased temperatures on our host–parasite system, we used two temperatures (20°C versus 24°C) and the presence/absence of the parasite in a fully factorial design. This resulted in four treatment combinations: 20°C and 24°C in the presence of the parasite (n = 55 individuals per treatment), and 20°C and 24°C in the absence of the parasite (n = 45 individuals per treatment). Previous work has shown that 20°C is a favourable thermal condition for optimal growth and reproduction in another species of Daphnia [39,40]. Prior studies on D. dentifera have also used a similar range of temperatures, since these represent ecologically relevant temperatures that are within the thermal tolerance of this species (which cannot be cultured in temperatures of 27°C or higher; [9,38]). We used 4°C increase in temperature since it reflects the average temperature change projected under global warming by the end of this century [41], with summer lake surface temperatures having risen at a rate of 0.34°C per decade since 1985 [42]. We used a larger number of animals in the parasite present treatment to accommodate potential mortality rate associated with spore checking at initial infection stage, although this ended up being low (see below). We collected neonates aged 1–2 days old from our laboratory stock and reared them at either 20°C or 24°C, maintaining each temperature treatment in a separate incubator (I-41VL, Percival Scientific). Since only two incubators were used due to logistical constraints, this meant that temperature treatment is confounded with incubator (as is common in the thermal biology literature). Each juvenile was kept individually in a 50 ml beaker filled with 50 ml filtered lake water and fed three times a week (20 000 cells ml⁻¹ A. falcatus). For the parasite-exposed treatment, we exposed juveniles that were 5–6 days old to parasite spores; control treatment animals were left unexposed.

To control for variation in developmental stage due to temperature differences during parasite exposure (i.e. faster growth at 24°C), we used a degree-day approach (accumulated product of time and temperature [43,44]) to expose Daphnia to parasites at an age of 5 and 6 days, for 24°C and 20°C, respectively. This allowed us to minimize potential body size differences between temperature treatments (χ² = 2.69, d.f. = 1, p = 0.101) during parasite exposure, since larger body size leads to higher spore intake [45]. Therefore, at degree-day 120 (i.e. 5 days × 24°C or 6 days × 20°C), we inoculated experimental animals with a suspension of spores obtained by crushing 45 D. dentifera infected with M. bicuspidata. Using a haemocytometer, we determined the spore density used (145 spores ml⁻¹), which is biologically relevant to field populations [12]. Meanwhile, the unexposed animals received a placebo solution containing the same amount of tissue from uninfected D. dentifera. Animals were exposed to parasites or placebo for 24 h; these animals were fed 20 000 cells ml⁻¹ A. falcatus and kept at 16:8 light:dark cycle. Experimental animals were then transferred individually to beakers filled with 50 ml filtered lake water, fed three times a week (20 000 cells/mL A. falcatus) and maintained at 16 : 8 light : dark cycle until the end of the experiment. The experiment was conducted in two temporal blocks between January and March 2022.

(c) . Data collection

(i) . Data collected only on hosts that were exposed to parasites

To determine the impact of temperature on the earliest stages of the infection process, we examined the hosts at the end of the 24 h inoculation period; in a prior study, the majority (75%) of ingested spores made it to the body cavity within the first 24 h [46]. Hosts were examined under an Olympus BX53F compound microscope at 200–400× magnification. We scanned for the presence of spores in the anterior and posterior regions of the host gut and body cavity, where spores are most likely to penetrate [12,46] (see also figure 1). Ingested spores were categorized into two classes (sensu [12]): embedded spores (i.e. spores partially embedded in the gut epithelium) and haemocoel spores (i.e. spores that had successfully penetrated into the body cavity). We were interested in quantifying the extent to which the gut is a barrier to infecting spores; we determined ‘gut resistance’ (sensu [28]) as the proportion of attacking spores (embedded spores + haemocoel spores) that were blocked by the gut barrier (i.e. embedded spores divided by attacking spores). To quantify the cellular immune response, we counted the total number of haemocytes attached to haemocoel spores (figure 1b). We also determined the number of haemocytes per spore by dividing the total number of haemocytes by the number of haemocoel spores.

As we quantified the cellular immune responses in infected individuals, we also imaged the gut epithelium at high magnification (400X) and measured the height of anterior midgut epithelial cells at the 90-degree bends in the C-shaped gut (figure 1), from where we determined the average gut epithelium thickness by haphazardly selecting three epithelial cells; the gut epithelium in Daphnia is one cell layer thick. We also measured body size by drawing a straight line from the center of the eye to the base of the spine. All measurements were made using the cellSens Standard Software (Olympus, version 1.18). In total, 50 and 53 individuals for control and warmer temperatures were successfully measured.

(ii) . Data collected on all hosts

To determine fitness components of the host, we checked all individuals daily for mortality and counted the number of offspring produced, which were then removed from the beakers. We estimated fecundity (total number of offspring) for all animals 20 days after exposure, when the last infected host was found dead. Upon death, each infected host was placed in a 1.5 ml tube of 100 µl deionized water and stored in a refrigerator for further spore yield counts. To determine spore yield per host, we ground the host using an electric pestle for 60 s and mixed the spore solution well before adding a 10 µl sample to a Neubauer Haemocytometer. We estimated the spore yield per infected individual by averaging the number of mature spores from four grids. We excluded animals that died within 7 days after exposure due to early mortality, during which infection status cannot be accurately diagnosed, resulting in a total of 162 individuals (control temperatures: n = 37 and 38 for control and parasite treatment; warmer temperatures: n = 44 and 43 for control and parasite treatment).

(d) . Statistical analyses

Most of the analyses focused on the treatments that were exposed to parasites, looking at the effect of temperature on key host traits and infection outcomes. These analyses used all of the hosts that were exposed to parasites. The only analyses that also included the unexposed animals were those looking at host fitness components. As described more below, for analyses of host fitness, the hosts that were exposed to parasites were divided into two groups—those that developed terminal infections and those that did not develop terminal infections—prior to analysis.

We analysed the data using generalized linear mixed models (GLMM) with the glmer function in the lme4 package [47] in R v. 4.1.2 [48]. Analysis of variance (ANOVA) was performed in the car package [49], with type II sums-of-squares. Model selection was conducted with a stepwise regression approach based on the Akaike information criterion by removing non-significant interactions. Once a significant interaction term was detected, Tukey post-hoc comparisons were made to assess differences among individual treatments in the emmeans package [50]. For all of the models described below, block was included as a random effect; treating block as a fixed effect did not affect our qualitative results. In addition, we centred the predictor variable gut epithelium thickness around zero prior to fitting the models.

To understand the effects of temperature on gut epithelium thickness, we included temperature treatment and body size as fixed effects with a Gaussian distribution. To investigate if guts were a barrier to spore infection, we analysed gut resistance (i.e. proportion of attacking spores that were blocked by the gut barrier) with a Gaussian distribution by including two fixed effects: temperature treatment and gut epithelium thickness, which was better fitted as a second-degree polynomial term. We analysed the number of haemocoel spores in a similar manner, including temperature treatment and gut epithelium thickness as fixed effects, but with a Poisson distribution. Gut resistance and the number of haemocoel spores are related to each other and generally negatively correlated, but also somewhat distinct. For example, a host that had a high feeding rate and relatively high gut resistance might still have a relatively high number of haemocoel spores.

To analyse the effect of temperature on cellular immune response, we analysed the total number of haemocytes with a negative binomial GLMM to account for data overdispersion. We included temperature treatment and the number of haemocoel spores as fixed effects. We further analysed the effect of temperature treatment on the number of haemocytes per spore with a Gaussian distribution. In addition, we included gut epithelium thickness as a fixed effect to explore potential trade-offs between cellular immune responses and physical barrier.

We were interested in how temperature and host traits jointly affected the probability of terminal infection—that is, the likelihood that an individual that was exposed to spores would develop an infection that produced new spores. Therefore, we modelled whether a host developed a terminal infection with a binomial distribution (terminal infection: 1; no terminal infection: 0) and logit link function. This analysis included temperature treatment, gut epithelium thickness and number of haemocytes per spore as fixed effects. Because of significant interactions between temperature treatment and gut epithelium thickness, we further analysed each temperature treatment separately, following common statistical practice [51], to evaluate differential effects of gut epithelium thickness on parasite infectivity and spore yield per host. We used a similar model to analyse effects of temperature and host traits on spore yield per host; for this response variable, the data were log transformed prior to analysis with a Gaussian distribution to meet the assumption of normality for regressions.

We also investigated the impact of temperature and infection on host fitness, quantified in terms of survival and fecundity. To do this, we analysed host survival with a Cox proportional hazard mixed effect model in the coxme package [52]. We included temperature (control/warmer) and parasite treatment (control/exposed/infected) as fixed effects and included block as a random effect. We used three categories related to parasite treatment to evaluate whether hosts that were exposed to parasites but that did not develop terminal infections (referred to as ‘exposed’ in our analyses) differed from uninfected controls and/or from hosts that were exposed and developed terminal infections (referred to as ‘infected’ in our analyses). The data were censored to indicate Daphnia that remained alive at the end of the study. Fecundity was analysed with a negative binomial GLMM to account for data overdispersion, with temperature and parasite treatments included as fixed effects.

3. Results

(a) . The impact of temperature on physical and cellular defense

Daphnia that experienced warmer temperatures had gut epithelial cells that were, on average, 16% thicker than those at control temperatures (control: 18.78 ± 0.46 µm, warmer: 21.76 ± 0.37 µm; figure 2a), even when controlling for body size (temperature treatment: χ² = 5.55, d.f. = 1, p = 0.019; figure 2b). Both thinner and thicker gut epithelia acted as barrier resistance to attacking spores—that is, a smaller proportion of attacking spores were successfully blocked in individuals with intermediate gut epithelium thickness (table 1; figure 2c). This led to a larger number of haemocoel spores in hosts with intermediate gut thickness (table 1; figure 2d). This relationship between gut epithelium thickness and gut resistance or haemocoel spores was consistent for the two temperature treatments (figure 2c,d), as indicated by the lack of statistically significant interaction terms between the temperature and gut epithelium thickness (table 1).

Figure 2.

The effects of temperatures on two different forms of defense (i.e. barrier and immunity) and their relationship with each other. (a) The gut epithelium, which serves as a barrier to parasite entry, is thicker in animals that experienced warmer temperatures, (b) even in animals of the same body size. (c) Animals with moderately thick guts had the lowest gut resistance, regardless of temperature, leading to (d) the number of haemocoel spores being highest for intermediate gut epithelium thickness; ‘gut resistance’ was quantified by dividing the number of embedded spores by the total number of attacking spores (that is, by the number of embedded spores + haemocoel spores), so the inverse relationship between gut resistance and haemocoel spores is expected. (e) Haemocoel spore number positively predicted total number of recruited haemocytes, with control temperatures consistently having more haemocytes than warmer temperatures. (f) The relationship between gut epithelium thickness and cellular immune responses, measured as the number of haemocytes per haemocoel spore, differed between the two temperatures. The box plot in (a) shows median values, the 25th and 75th percentiles, and interquartile ranges. Solid and dashed lines indicate regressions for control and warmer treatments, respectively, predicted from GLMMs. Shaded regions represent 95% confidence intervals. Each panel also plots the raw data points (each from a single animal), with rug plot shown in (c). (Online version in colour.)

Table 1.

Results from the final models analysing the traits associated with infection processes. Statistically significant p values are highlighted in bold.

dependent variable	explanatory variables	χ2	d.f.	p value
gut resistance	gut thickness	0.73	1	0.394
	gut thickness2	4.44	1	0.035
	temperature	0.002	1	0.962
	gut thickness × temperature	1.06	1	0.303
	gut thickness2 × temperature	0.87	1	0.350
haemocoel spore	gut thickness	2.19	1	0.139
	gut thickness2	4.58	1	0.032
	temperature	0.09	1	0.770
	gut thickness × temperature	0.003	1	0.954
	gut thickness2 × temperature	0.43	1	0.513
total recruited haemocytes	haemocoel spore	44.88	1	<0.001
	temperature	16.64	1	<0.001
	haemocoel spore × temperature	2.13	1	0.145
haemocytes per spore	gut thickness	0.26	1	0.607
	gut thickness2	0.06	1	0.805
	temperature	9.09	1	0.003
	gut thickness × temperature	1.63	1	0.201
	gut thickness2 × temperature	0.14	1	0.707

Turning to cellular immune responses, we found that both the number of haemocoel spores and temperature impacted the number of haemocytes that attacked spores (table 1, figure 2e): as more spores penetrated into the body cavity, the total number of haemocytes that attached to the infecting spores increased, and hosts reared at warmer temperatures had consistently reduced haemocyte recruitment. There was no significant relationship between gut thickness and haemocytes per spore, but, consistent with the pattern for total haemocyte recruitment, there were fewer haemocytes per spore at warmer temperatures (table 1; figure 2f).

(b) . The impact of temperature on host and parasite fitness

Probability of terminal infection was predicted by gut thickness, but the patterns differed markedly between temperature treatments (table 2; figure 3a). Specifically, the probability of terminal infection increased as gut epithelium thickness increased at control temperatures (χ² = 4.74, d.f. = 1, p = 0.029), whereas the probability of terminal infection decreased as gut epithelium thickness increased at warmer temperatures (χ² = 4.08, d.f. = 1, p = 0.043). While Daphnia with thicker guts had similarly high probability of terminal infection irrespective of temperature treatment (χ² = 0.70, d.f. = 1, p = 0.401; figure 3b), those with thinner guts tended to have relatively lower probability of terminal infection at control than warmer temperatures (χ² = 5.50, d.f. = 1, p = 0.019; figure 3b). Haemocytes per spore did not predict the likelihood of terminal infection in Daphnia with thicker guts (χ² = 0.16, d.f. = 1, p = 0.693) or those with thinner guts (χ² = 1.96, d.f. = 1, p = 0.162).

Table 2.

Results from the final models analysing the parasite fitness components. Statistically significant p values are highlighted in bold.

dependent variable	explanatory variables	χ2	d.f.	p value
probability of terminal infection	gut thickness	0.0008	1	0.977
	temperature	1.08	1	0.299
	haemocytes per spore	3.01	1	0.083
	gut thickness × temperature	6.68	1	0.010
	haemocytes per spore × temperature	0.16	1	0.689
spore yield per host	gut thickness	0.04	1	0.833
	temperature	2.23	1	0.135
	haemocytes per spore	3.08	1	0.079
	gut thickness × temperature	6.27	1	0.012
	haemocytes per spore × temperature	0.56	1	0.453

Figure 3.

(a,b) The probability of terminal infection and (c,d) spore yield per host in relation to gut epithelium thickness at control and warmer temperatures. Significant interactions between temperature treatments and gut epithelium thickness were detected, and so the data were split into two parts in (b,d), based on the intersection values of gut epithelium thickness in (a,c) to evaluate the effects of temperature. Data for all hosts that were exposed to parasites are shown, with rug plot shown in (a). Solid and dashed lines indicate regressions for control and warmer treatments, respectively, predicted from GLMMs. Means and standard error bars are shown. (Online version in colour.)

When considering spore yield, the pattern differed depending on whether the analysis included all hosts that were exposed to the parasite, or if it just included those that developed terminal infection. For Daphnia that developed terminal infection, neither gut thickness (χ² = 0.08, d.f. = 1, p = 0.779) nor temperature (χ² = 1.85, d.f. = 1, p = 0.173) impacted spore production. When considering all hosts that had been exposed to parasites, there was an effect of gut thickness and temperature on spore yield (table 2; figure 3c). Of all animals that were exposed, more spores were produced in hosts with thicker guts only at control temperatures (χ² = 5.93, d.f. = 1, p = 0.015) but the relationship was inversed at warmer temperatures (χ² = 5.19, d.f. = 1, p = 0.023; figure 3c). Daphnia with thicker guts had similarly high numbers of spores produced in both control and warmer temperatures (χ² = 1.04, d.f. = 1, p = 0.309), whereas Daphnia with thinner guts produced fewer spores at control than at warmer temperatures (χ² = 5.99, d.f. = 1, p = 0.014; figure 3d). Haemocytes per spore did not predict spore yield in Daphnia with thicker guts (χ² = 0.01, d.f. = 1, p = 0.917) or those with thinner guts (χ² = 1.90, d.f. = 1, p = 0.168). Thus, the most complete measure of parasite fitness in our study—the spore yield per exposed host—was impacted by gut thickness and temperature, but that effect was due to variation in the likelihood of developing a terminal infection, rather than variation in spore yield in hosts that developed terminal infections.

Daphnia fitness was impacted by both parasite treatment and temperature (figure 4). Daphnia exposed to Metschnikowia suffered from reduced fecundity but the extent depended upon temperature treatments (temperature × parasite treatments: χ² = 20.76, d.f. = 2, p < 0.001; figure 4a). Specifically, Daphnia infected by Metschnikowia experienced fecundity reduction compared to unexposed Daphnia at control temperatures (compare the blue ‘control’ bar and the blue ‘infected’ bar in figure 4a; post-hoc comparison control v. infected: z = 8.11, p < 0.001), but there was greater reduction at warmer temperatures (red ‘control’ bar versus red ‘infected’ bar in figure 4a; post-hoc comparison control v. infected: z = 15.83, p < 0.001). As a result, warmer temperatures led to lower reproductive success of infected Daphnia compared to control temperatures (blue ‘infected’ bar versus red ‘infected’ bar in figure 4a; post-hoc comparison control v. warmer: z = 3.12, p = 0.002). For both temperature treatments, Daphnia that were exposed but uninfected had intermediate numbers of offspring—they produced more offspring than infected animals but fewer than the control (‘exposed’ treatment in figure 4a); no difference was found between the two temperature treatments (post-hoc comparison control vs warmer: z = −0.15, p = 0.878). Host survival was lower at warm than at control temperatures, and survival was highest in unexposed controls, followed by exposed but uninfected, and lowest in infected animals (figure 4b). Host survival was independently affected by temperature (χ² = 28.40, d.f. = 1, p < 0.001) and parasite treatments (χ² = 85.93, d.f. = 2, p < 0.001), with the lowest overall survival for animals that were infected at warm temperatures.

Figure 4.

The effect of temperature change and Metschnikowia infection on Daphnia fitness components, measured as (a) fecundity (i.e. total number of offspring) and (b) survival, i.e. the proportion of individuals remaining alive at each time. In parasite treatment, ‘control’ represents hosts unexposed to parasites, whereas ‘exposed’ and ‘infected’ represent hosts exposed to parasites but that remained uninfected or became infected, respectively. The box plots show median values, the 25th and 75th percentiles, and interquartile ranges. Solid lines represent the infected individuals, dashed lines represent exposed, and dotted lines represent uninfected individuals. (Online version in colour.)

4. Discussion

Our results showed that warmer temperatures altered physical barriers and cellular immune responses to infection and exacerbated the effects of the fungal parasite Metschnikowia bicuspidata on the host Daphnia dentifera. While hosts at warmer temperatures had (beneficial) thicker guts, they also had dampened cellular immune responses. At the same time, infected hosts suffered a greater reduction in fecundity and lifespan at warmer temperatures. Underlying these results may be a fundamental change in the effectiveness of each defense mechanism at warmer temperatures. In our study, the barrier defense became more robust at warmer temperatures, but the cellular immune response was diminished. Moreover, we found that the nature of the relationship between the barrier response trait—gut epithelium thickness—and the probability of developing terminal infections reversed at warmer temperatures. Hosts with thicker gut epithelia were more likely to develop terminal infections at control temperatures but less likely to develop terminal infections at warmer temperatures (figure 3). Therefore, it is clear that physical barriers are impacted by thermal change, and that the relationship between this key trait and infection is itself sensitive to changes in temperature.

Overall, our findings demonstrate that we must understand how temperature influences key traits and their relationship with infection outcomes to predict the ecological and evolutionary consequences of host–parasite interactions in a warmer world. Prior work on how temperature modulates host–parasite interactions and disease dynamics has centred on immune activity [53] or other rapidly changing traits such as feeding rates [54]. However, mechanical barriers—the first line of defense—are as important as other traits. For example, the gut epithelium in Daphnia has been widely found to function as a physical barrier to ingested parasites [11,12,55,56] (discussed further in the next paragraph), and physical barriers are also commonly found in a wide variety of systems, such as midgut barriers in mosquitoes [57] and cuticles in plants [32]. Nevertheless, while it is clear they are important, how environmental factors influence physical barriers to parasite infection remains unexplored. One reason why this is a problem is because physical barriers are likely less plastic—at least on short time scales—than the responses that have been considered, such as immune responses and behaviours such as feeding rate. Similarly, a physical barrier is likely to remain for longer than many other plastic responses once it has been induced. This difference in timescale and permanence might lead to qualitatively different responses of physical barriers to environmental changes such as warming temperatures. To our knowledge, our study provides the first evidence of a temperature-driven shift in a physical barrier against parasites. Moreover, our results suggest that not only can barrier defenses change in response to warmer temperatures, but that it is important to consider the potential that physical barriers might operate differently at warmer temperatures.

Previous work in this system also found that the gut epithelium acted as a barrier to M. bicuspidata infection, with thinner guts being more likely to prevent spore penetration [12]. Our findings corroborate this within the same range of gut epithelium thickness (10–20 µm). However, we found an overall U-shaped relationship between gut epithelium thickness; beyond approximately 20 µm, thicker guts were more resistant to attacking spores (figure 2c). Because of this nonlinear relationship between gut thickness and resistance to attacking spores, we found fewer haemocoel spores at both thinner and thicker ends of the gut epithelium gradient (figure 2d). This nonlinear relationship suggests that there might be distinct mechanisms driving the effectiveness of gut resistance to spores in relation to the range of gut epithelium thickness. It also has the potential to influence tradeoffs associated with infection defense. For example, it has been hypothesized that the penetrability of gut barrier is likely associated with the ability to absorb nutrients [12,58]. Thus, the unimodal relationship between gut resistance and gut thickness might also influence resource acquisition. The potential for tradeoffs between these key traits (resistance to infection and resource acquisition), and for the nature of that tradeoff to change at warmer temperatures, warrants further study.

Temperature did not influence the relationship between gut thickness and gut resistance (figure 2c), and while the immune response was dampened at warmer temperatures, temperature did not alter the relationship between gut thickness and haemocytes per spore (figure 2f). Given this, we would predict that there should not have been a significant effect of temperature on the relationship between gut thickness and the probability of terminal infection. Instead, there was a strong effect: thicker guts were associated with increased risk of terminal infection at control temperatures but decreased risk at warmer temperatures (figure 3a). This decreased risk of terminal infections associated with thicker guts at warmer temperatures is consistent with the gut resistance pattern, where thicker guts were more resistant. However, why individuals with thicker, more resistant guts were more likely to develop terminal infections at control temperatures is unclear and warrants further investigation.

In addition to barrier defenses, hosts employ immune defenses, including cellular immune responses [12,59]. By contrast to the pattern for barrier defenses, at warmer temperatures, the cellular immune response was reduced; hosts reared at warmer temperatures were not able to recruit as many haemocytes after spores penetrated the body cavity (figure 2e). Haemocytes are essential components of cellular immune responses in invertebrates, and in some cases, can prevent infection development [11]. However, in our study, the reduction in haemocyte number at warmer temperatures did not translate into greater likelihood of terminal infection. Instead, our results are consistent with accumulating evidence suggesting that increasing haemocyte number can be a symptom of infection, rather than an effective immune response [12,60]. Here, hosts with more haemocytes per spore did not have lower terminal infection likelihood or spore yield; instead, there was either no significant relationship or a slightly positive one between haemocytes per spore and terminal infection and spore yield (table 2; electronic supplementary material, figure S1). While this might suggest that the host immune response is not effective, other evidence is at odds with this: in this study and in others [28], not all hosts that had spores penetrate into the body cavity developed terminal infections, suggesting that there can be resistance even after a spore penetrates into the haemocoel. Further work aimed at uncovering immune responses beyond the initial infection stages will likely help explain some of the variation in terminal infections [61,62].

As expected given the virulence of this parasite, hosts that developed terminal infections suffered from shortened lifespan and fecundity loss. Nevertheless, at the moment there is no consensus on whether rising temperatures increase parasite virulence [63], and there is evidence that this will not always be the case [40]. In our study, the decreases in host fitness were particularly acute at warmer temperatures (figure 4). We found that 39.5% of the exposed hosts at control and 25.6% of the exposed hosts at warmer temperatures did not develop terminal infection, and that they outperformed those that developed terminal infections (figure 4). Nevertheless, at both control and warmer temperatures, Daphnia that were exposed to the parasite but that did not develop terminal infections still performed worse than those that had not encountered parasites, suggesting that fighting infection is costly, as has been found in an earlier study [64]. Most, but not all, of the ‘exposed’ animals in our study were observed to have haemocoel spores 24 h after the end of the exposure period and, therefore, successfully fought off early infections.

In systems where immune responses are largely modulated by environmental temperature (e.g. ectotherms), warmer temperature can interact with immune functions in complex and unpredictable ways [65]. Temperature increase is known to cause variation in the expression of immune responses in both terrestrial and aquatic systems [11,66], and developing a mechanistic understanding of how temperature modulates host immune responses and host–parasite interactions has become a central focus in ecological immunity [65,67,68]. Our results demonstrate that temperature modulates the infection process and the outcome of host–parasite interactions in complex ways. Moreover, our results demonstrate the importance of considering barrier traits in addition to traits such as immune responses. Better understanding of host–parasite interactions and disease dynamics in a changing world will require understanding how temperature affects key host susceptibility traits, including barrier defenses, as well as the relationship between those traits and infection outcomes.

Data accessibility

Data are provided as an Excel file in the electronic supplementary material [69].

Authors' contributions

S.-J.S.: conceptualization, data curation, formal analysis, investigation, methodology, project administration, resources, validation, visualization, writing—original draft, writing—review and editing; M.K.D.: conceptualization, data curation, formal analysis, investigation, methodology, project administration, resources, validation, writing—original draft, writing—review and editing; R.N.J.: data curation, investigation, methodology, writing—original draft, writing—review and editing; M.A.D.: conceptualization, formal analysis, funding acquisition, investigation, methodology, project administration, resources, supervision, validation, visualization, writing—original draft, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

S.-J. S. was supported by National Science and Technology Council, Taiwan (grant no. 111-2628-B-002-050-) and Ministry of Education, Taiwan (grant no. 111V1024-1). This work was supported by the Gordon and Betty Moore Foundation (grant no. GBMF9202; https://doi.org/10.37807/GBMF9202).