Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Jan 1.
Published in final edited form as: Ecol Lett. 2023 Dec 19;27(1):e14352. doi: 10.1111/ele.14352

Into the danger zone: how the within-host distribution of parasites controls virulence

Pieter T J Johnson 1,*, Tara Stewart Merrill 1,2, Dana M Calhoun 1, Travis McDevitt-Galles 1,3, Brendan Hobart 1
PMCID: PMC10872350  NIHMSID: NIHMS1951102  PMID: 38115188

Abstract

Despite the importance of virulence in epidemiological theory, the relative contributions of host and parasite to virulence outcomes remain poorly understood. Here, we use reciprocal cross experiments to disentangle the influence of host and parasite on core virulence components – infection and pathology – and understand dramatic differences in parasite-induced malformations in California amphibians. Surveys across 319 populations revealed that amphibians’ malformation risk was 2.7x greater in low-elevation ponds, even while controlling for trematode infection load. Factorial experiments revealed that parasites from low-elevation sites exhibited higher per-parasite pathogenicity (reduced host survival and growth), whereas there were no effects of host source on resistance or tolerance. Parasite populations also exhibited marked differences in within-host distribution: ~90% of low-elevation cysts aggregated around the hind limbs, relative to <60% from high-elevation. This offers a novel, mechanistic basis for regional variation in parasite-induced malformations while promoting a framework for partitioning host and parasite contributions to virulence.

Keywords: host-parasite interaction, virulence evolution, disease ecology, global change, parasite manipulation, emerging disease, trematode, Ribeiroia

Graphical Abstract

Despite the importance of virulence in epidemiological theory, the relative contributions of host and parasite to virulence outcomes remain poorly understood. Here, we use reciprocal cross experiments to understand host and parasite contributions to parasite-induced malformations in amphibians. Field and experimental studies reveal that dramatic differences in virulence among population owes to variation in where trematodes infect their hosts, rather than host defenses, offering a framework to mechanistically partition virulence.

Introduction

Although virulence is often treated as a property of the parasite, growing evidence highlights the roles of both host and parasite processes in controlling virulence outcomes (e.g., Little et al. 2010). Virulence, defined as the consequences of infection for host fitness, is the product of two key processes: the amount of infection in a host (parasite infection load or intensity) and the amount of damage caused per parasite or unit of infection. Both processes have the potential to include contributions from the host, the parasite, and their interaction (including environmental effects). For instance, the amount of infection in a host (conditional on exposure) can be limited by host defenses such as anti-parasite behaviors, immunological resistance, and post-infection clearance (Schmid-Hempel 2009; Gibson & Amoroso 2022). We treat these as components of “host resistance” (Stewart Merrill & Johnson 2020). Reciprocally, parasites may have “offensive” traits that enhance their ability to infect, persist, or replicate within hosts, which can be considered as contributions to “parasite infectivity” (also called “exploitation”). Thus, the likelihood a host is infected as well as parasite load represents the dynamic interplay between parasite infectivity and host resistance (see Fig. 1).

Figure 1.

Figure 1.

Open in a new tab

Partitioning out the joint contributions of hosts and parasites on virulence. Virulence (loss of fitness among infected hosts) is the result of two processes: infection (number or load of parasites relative to exposure) and host damage (amount of damage in the host for each unit of parasite infection). Each process includes host and parasite constituent components that can be statistically decomposed using an experimental approach. Following exposure of a host of type i to a known quantity/dosage of parasites of type j, the resulting parasite load will be a function of parasite infectivity and host resistance. If different host and parasite types are factorially crossed across a dosage range, infection load (I) can be modelled as Iij = (ci − di)*Ej, where Ej is exposure, ci is parasite infectivity, and dj is host resistance. Similarly, the pathology resulting in infected hosts will be a function of parasite load (I), host tolerance (ei), and the per-parasite pathogenicity (fi). Thus, fitness (W) of host type i exposed to parasite type j is Wij = aij + (ei − fi)*Iij. Note that the intercept, aij, can vary as a function of (1) host population vigor (i.e., fitness in the absence of infection; see the far-right plot) and/or (2) ‘costs of exposure’ that occur independent of whether successful infection occurs. The contributions of vigor and ‘cost of exposure’ to variability in aij can be quantified by comparing fitness of unexposed hosts to hosts that were exposed but did not support any successful infections (see Rohr et al. 2010). Tolerance and pathogenicity can be detected as statistically significant interaction terms with parasite load, or visually as different slopes in the relationship between parasite load and host fitness as a function of host or parasite population source/genotype. In this theoretical example, colors represent different parasite populations and line types (dashed or solid) represent host populations.

The second process mediating virulence involves the amount of pathology the host experiences, conditional on parasite load or intensity. Stated another way, how much host damage (fitness loss) is associated with each unit increase in infection? How this is measured may depend on whether parasites can replicate within hosts, such as for bacteria and viruses, or whether each parasite represents an independent infection event, as for many macroparasites. Once again, the outcome of this interaction can include both host and parasite contributions. Host tolerance is defined as the evolutionary or environmental capacity of hosts to limit the damage caused by invading parasites, which is recognized as a distinct strategy from host resistance (Råberg et al. 2009; Medzhitov et al. 2012; Tadiri et al. 2021). By extension, parasites can also vary in per-parasite pathogenicity, or the amount of damage caused per unit increase in infection, which emerges as different slopes in the relationship between parasite load and pathology among parasite genotypes, while controlling for host genotype or condition (Fig. 1) (Råberg 2014). For instance, Gonzales et al. (2022) showed that a plant virus experimentally evolved under simulated drought conditions was substantially less pathogenic to its plant host compared with a viral strain that evolved under moderate climate, despite no differences in viral multiplication rates.

Identifying the respective contributions of hosts and parasites to virulence is essential to efforts aimed at understanding and managing disease outcomes. In an influential contribution, Råberg et al. (2009) outlined the methodological approaches and statistical framework necessary to quantify host resistance and host tolerance, prompting subsequent studies that have measured these dual components of host defense across host genotypes, species, and developmental stages (e.g., see Rohr et al. 2010; Knutie et al. 2017; Klemme et al. 2020). Here we emphasize the parallel components of parasite offense (infectivity and pathogenicity). Disentangling the relative contributions of both host and parasite to virulence requires an experimental approach that factorially crosses host and parasite genotypes while also varying parasite exposure or load (see Råberg 2014). Differences in host resistance and parasite infectivity appear as mean differences in parasite load for a given dosage between host or parasite populations, respectively. Likewise, statistical interactions between parasite load and host or parasite genotype in determining host fitness responses are indicative of differences in host tolerance or parasite pathogenicity, respectively. Additional interactions between host and parasite source would suggest that outcomes are the joint product of both. Although such approaches have featured prominently in disease-related studies of local adaptation (e.g., Greischar & Koskella 2007), we are aware of no studies that have extended this framework to also disentangle host and parasite contributions to virulence (i.e., host resistance and tolerance as well as parasite infectivity and pathogenicity).

An important additional consideration in the study of virulence is the specific locations where parasites are found within their hosts. Many parasites exhibit a high degree of specificity with respect to the organs or tissues they infect, owing to metabolic requirements, accessibility, life cycle, or host defenses (Holmes 1973). But infection location is not always static, and even small-scale differences in where parasites colonize can sharply influence host pathology. For example, the trematode Curtuteria australis infects the foot of marine cockles to form a resting stage (metacercaria), ultimately waiting for the cockle to be consumed by predatory birds (oystercatchers). When parasites invade the tip of the foot (rather than the base), they impair the ability of the host to burrow and thereby increase its risk of predation (Thomas & Poulin 1998). Leung et al. (2010) used an experimental approach to show that parasites make this choice based on the number of metacercariae already present in the foot tip, rather than any genetic predisposition. Where parasites occur in the host can also be influenced by host factors. Sears et al. (2013) showed that while plagiorchiid trematode cercariae preferred to infect the heads of their tadpole hosts, tadpoles used behavioral defenses to deflect invading parasites toward their tails where they had fewer effects on host growth. These studies emphasize that, alongside considerations of parasite load, careful attention should be paid to the within-host distribution of infection.

A particularly striking example of parasite-induced virulence involves the trematode Ribeiroia ondatrae, which induces mortality and severe limb malformations in amphibians. Affected amphibians suffer reduced ability to swim, jump, and capture prey, and malformed frogs rarely survive to sexual maturity (Goodman & Johnson 2011; Lunde et al. 2012). Based on experimental studies and comparative field surveys, the frequency of malformations increases directly in response to average parasite load, in some cases affecting over 50% of recently-metamorphosed frogs (Johnson et al. 2012, 2013; Lunde et al. 2012). Intriguingly, however, long-term study of interactions between amphibians and R. ondatrae reveals sharp differences in the relationship between parasite load and pathology among geographic regions (McDevitt-Galles et al. 2020). In low-elevation populations of Pacific chorus frogs (Pseudacris regilla), malformation frequency increases steeply with average R. ondatrae load, whereas this relationship is more moderate among populations in the Cascade and Sierra Nevada mountains (>1500 m ASL) (Fig. 2). The parasites from each region also exhibit conspicuous differences in their distribution within infected frogs: while ~90% of encysted R. ondatrae from low-elevation frogs are aggregated around the hind limbs (where malformations primarily occur), parasites from high-elevation sites show much weaker within-host aggregation. These observations indicate that both the distribution of parasites within amphibian hosts and the severity of pathology resulting from such host–parasite interactions differ markedly between regions. Less clear, however, is whether these differences are linked or how they stem from characteristics of the host, the parasite, or their interactions, for which experimental testing is necessary.

Figure 2.

Figure 2.

Open in a new tab

The relationship between parasite infection load and host malformations (virulence) between low- and high-elevation ponds. (A) Each data point represents a pond-by-year sampling event of recently metamorphosed Pacific chorus frogs (Pseudacris regilla) (n = 319), for which we quantified both the average number of Ribeiroia ondatrae metacercariae per frog and the proportion of frogs exhibiting limb abnormalities. The best-fit lines and standard error intervals from a binomial GLMM are depicted with teal solid lines for low-elevation ponds (<1000 m ASL) and with peach dashed lines for high-elevation ponds (>1500 m ASL). (B) Center panels illustrate representative habitats for high- and low-elevation ponds. (C) Distribution of sampled high- and low-elevation ponds in California. Low-elevation study sites (teal points) are located in the Bay Area of California whereas high-elevation sites (peach points) are located across the lower Cascade and High Sierra Mountain ranges. California elevation raster was downloaded from Natural Earth at 1:10 million scale (https://www.naturalearthdata.com). High-elevation site locations are jittered to improve clarity. Note that the low-elevation sites are spatially separated from high-elevation sites by approximately 250-300 km.

To investigate the mechanisms underlying virulence differences among amphibian host populations, we combined long-term and large-scale field data with factorially crossed experiments involving host source, parasite source, and exposure dosage. By studying R. ondatrae infection and the resultant virulence outcome (malformations) from 319 amphibian populations, we first illustrate strong differences in the relationship between parasite load and pathology as a function of elevation. Detailed information on within-host parasite distributions led us to hypothesize that increased malformations at low elevation resulted from greater aggregation of encysted parasites around the hind limbs. To understand why such differences occurred and the joint contributions of host and parasite processes, we reciprocally crossed hosts and parasites from high- and low-elevation sites using a reaction norm design with four levels of parasite exposure (16 treatments). Response variables included: i) parasite distribution within hosts, which we hypothesized was the basis for the differences in virulence, ii) parasite infection success, which allowed us to partition host resistance from parasite infectivity, iii) and host pathology (mortality and body size following exposure), which allowed us to differentiate host tolerance from parasite pathogenicity. An integrated framework for decomposing host and parasite contributions to infection outcomes has the potential to help understand the process through which pathology occurs and shed light on its evolutionary origins.

Materials and Methods

Field study: assessing natural variation in virulence and within-host parasite distribution

Between 2015 and 2021, we surveyed 319 populations of P. regilla from lentic habitats in California. Sites spanned an elevation gradient, with low-elevation sites in the Bay Area (100 to 820 m ASL) and higher elevation sites (1610 to 2432 m) in the Cascade and Sierra Nevada mountain ranges (Fig. 2). Low- and high-elevation sites are also geographically separated by ~250 km. At each pond, we inspected late-stage larvae or recently metamorphosed P. regilla to determine the frequency of limb malformations (one metric for virulence) following standard protocols (see Johnson et al. 2002). Captured individuals were released after examination, except for a random subset that was systematically necropsied to quantify infection by the trematode Ribeiroia ondatrae. This parasite has a complex life cycle in which infective cercariae emerge from freshwater snails and penetrate the tissue of larval amphibians to form an encysted stage (metacercaria) (Johnson et al. 2004). When infections occur during tadpole limb development, they can induce limb malformations hypothesized to increase predation by water birds, which are the parasite’s definitive hosts (Goodman & Johnson 2011; Johnson et al. 2011). Metacercariae in amphibians occur just below the skin around the base of the hind limbs, near the area of tail resorption, or in the head around the lower mandible. To identify within-host parasite distributions, we recorded the location of each encysted metacercaria as in the ‘hind limb region’ (cysts posterior to the mid-abdomen) or ‘anterior region’ (cysts anterior to the mid-abdomen). For a subset of hosts, we created heat maps to visually illustrate the within-host distributions of R. ondatrae infection (see SI1).

Reciprocal cross infections: partitioning virulence into host and parasite sources

To understand how parasite source, host source, and their interaction affected parasite distribution and host pathology, we conducted a reciprocal cross infection experiment involving Pacific chorus frog (P. regilla) hosts and R. ondatrae cercariae. The study was a 2 x 2 x 4 factorial design (16 treatments), with variation in host source (high and low elevation), parasite source (high and low elevation) and parasite dosage (4 levels). Because pathology from trematode infections is load dependent, the range in exposure dosages was a key component for testing differences in the parasite load–pathology relationship as a function of host and parasite source populations (see Fig. 1). Each treatment included between 14 and 30 replicates for a total of 342 experimental animals.

Lab-reared P. regilla from low- and high-elevation sites were exposed during early limb development to cercariae collected from snails originating from high- and low-elevation source ponds using standard protocols (see Stewart Merrill et al. 2022). Tadpoles were maintained individually in 1.5 L containers until 20 days post-exposure, at which point they were euthanized, measured (snout–vent length), and dissected. We recorded the number of encysted metacercariae and their position within the host, with an emphasis on the hind limb region versus anterior infections. Although we intended to measure limb malformations, COVID-related restrictions prevented us from raising animals to metamorphosis. Our analyses of pathology therefore focus on host survival and body size (as a proxy for growth).

Testing parasite source population effects in a second host species

Because our preliminary results indicated that differences in distribution and virulence were driven by parasite source population, rather than variation in the host, we exposed a second amphibian host species to parasites from high- and low-elevation source populations to further investigate the generality of this pattern. We obtained eggs of the western toad (Anaxyrus boreas) from a site in Oregon (Lake Penhollow) and raised tadpoles to Gosner (1960) stage 28–30. Tadpoles were individually exposed to 10 R. ondatrae cercariae obtained from either a low-elevation or high-elevation source pond following the protocol described above. Although our chorus frog experiment used a range of doses, we limited exposure in toads to 10 cercariae because higher doses can result in high mortality (Stewart Merrill et al. 2022). Each treatment was replicated 20 times for a total of 40 animals. After 36 hours, all tadpoles were euthanized, measured, and dissected to quantify the number and location of metacercariae.

Statistical analysis

We analyzed the relationship between parasite load and the frequency of limb malformations in field-collected P. regilla using a generalized linear mixed model with a binomial error distribution. The cbind function in R (R Core Team 2022) was used to combine columns for the number of malformed and normal frogs, such that each row represented a site-by-year combination weighted by the number of examined frogs. As fixed effects, we included parasite infection load (average R. ondatrae per P. regilla) and a factor to represent whether a site was “low elevation” (<1000 m) or “high elevation” (>1500 m), which marked a natural division in the dataset (see Fig. S1), and the interaction between these variables. Elevation values were obtained using the elevatr package in R (Hollister et al. 2021). To account for sources of autocorrelation, we incorporated random intercept terms for site, year, and the site-by-year combination (i.e., an observation-level random effect to help address overdispersion). Average host body size at each site was included as a covariate. Only site-year observations with at least 15 frogs examined for malformations and five or more frogs dissected to quantify infection were included.

We also used a binomial GLMM to test how pond elevation (high vs. low) influenced the fraction of of metacercariae in the hind limb region versus the anterior region for each frog (using cbind to combine columns for the number of parasites in each area) (n = 1,678 frogs). Because changes in parasite distribution could also be driven by total parasite load, particularly if space within a host becomes limiting, we included total parasite load and its interaction with elevation as fixed effects. Host size (snout–vent length) was included as an additional covariate to capture potential variation in either resource availability or time-since-metamorphosis, either of which could affect host tolerance (including differential mortality between malformed versus normal frogs). Random intercept terms were included for site, year, and host.

For analyses of experimental infections, our aim was to identify the relative contributions of host and parasite processes to parasite distribution, infection success, and host pathology. We generally included fixed effects for host source (high vs. low), parasite source (high vs. low), parasite load, and all pairwise interactions. Host size or developmental stage (Gosner 1960) was treated as a covariate while host identity was included as an observation-level random effect. Analyses of parasite distribution utilized a binomial GLMM to contrast the number of parasites from the hind limb region to that in the anterior region, for which control hosts were omitted. Among hosts that survived for at least 10 days, we analyzed parasite load (infection success) using an overdispersed Poisson GLMM with exposure dosage as an additional fixed effect. To investigate host pathology, we focused on interactions between parasite load and host or parasite source population to derive insight into host tolerance and per-parasite pathogenicity. We used a Gaussian GLM to analyze host body size and a binomial GLMM for host survival. Control (unexposed) hosts were included in the analysis to estimate differences in baseline host survival independent of infection (“host vigor”). However, because control hosts were not exposed to parasites, they could not be assigned a value for “parasite source”, which is an important model term for hypothesis testing. We therefore listed each control animal twice in the dataset, once with a designation of ‘low-elevation’ parasite source and once as ‘high-elevation’, and used a host-level random effect to identify these entries as stemming from the same individual. This constrained the average survival of uninfected hosts with different parasite sources to be the same, such that any differences in host or parasite populations would emerge from the fitted slopes (see also Råberg et al. 2007).

Models were implemented in R (4.2.1) using the glmmTMB package (Brooks et al. 2017). All numeric terms were centered and scaled prior to inclusion. We used likelihood ratio tests to progressively simplify models by removing the least influential terms (beginning with interactions) until all terms were considered significant (P < 0.05).

Results

Field study: assessing natural variation in within-host parasite distribution and virulence

Ribeiroia ondatrae infection load was a strong, positive predictor of malformation frequency, which varied from 0 to 75% among sites (Binomial GLMM; βload = 0.544 ± 0.128, P < 0.0001; n = 319) (Fig. 2). Parasite load also interacted with elevation (βload-by-elevation[low] = 0.986 ± 0.215, P < 0.0001), such that the effect of infection on malformation frequency was 2.7x stronger at low-elevation ponds compared with high-elevation sites (Fig. 2). There was no added influence of host body size, and results were comparable if elevation was treated as a continuous variable rather than a factor. This relationship persisted even when analyzing only the low-elevation sites with elevation treated as a continuous variable (βload-by-elevation = −0.641 ± 0.246, P < 0.01; n = 293).

Parasites from low and high elevation sites exhibited strikingly different distributions within infected hosts. Among the 1,678 frogs in which the within-host location of each R. ondatrae was recorded, elevation negatively predicted the fraction of metacercariae occurring around the hind limb region (Binomial GLMM; βelevation[low] = 1.77 ± 0.404, P < 0.0001), with no effect of host body size, total parasite load, or its interaction with elevation (all P > 0.1). On average ± 1 SE, 86.6% ± 0.7% of parasites from low-elevation frogs were observed near the hind limbs, relative to 67.5% ± 1.9% among frogs from high elevation (Fig. 3).

Figure 3.

Figure 3.

Open in a new tab

Distribution of R. ondatrae parasites within amphibian hosts from low- and high-elevation ponds. Violin plots illustrate the proportion of parasites per host occurring in the hind limb region specifically, as opposed to those found more anterior. Dark bands represent the mean for examined P. regilla from high elevation and from low elevation (n = 1,678). High-resolution heat map visualizations of the within-host distribution of R. ondatrae metacercariae from high- and low-elevation P. regilla (n = 20 and 35, respectively). The relative size of dots and the heat color reflect the average number of detected metacercariae in a given grid cell (see SI1). Frog image by M. Benard, used with permission.

Reciprocal cross infections: partitioning virulence into host and parasite sources

Parasite source population – but not host population source or their interaction – influenced infection success and host pathology. The number of R. ondatrae per host increased with exposure dosage and was greater among hosts exposed to high-elevation parasites (Poisson GLMM; βdose = 1.11 ± 0.069, P < 0.0001; βparasite_low = −0.696 ± 0.158, P < 0.0001), with no effects of host source, host body size, or the host-by-parasite source interaction (all P > 0.1) (Fig. 4). Parasite load and parasite source further interacted to determine host mortality and growth. The scaled per-parasite effect on host mortality risk was ~1.8-fold greater for R. ondatrae from low-elevation ponds than for high-elevation parasites (Binomial GLMM; βload-by-parasite_low = 11.57 ± 3.69, P = 0.0017) (Fig. 4). Hosts collected from low elevation ponds also had lower overall mortality than those from high-elevation ponds (i.e., greater ‘general vigor’; βhost_low = −7.27 ± 3.66, P = 0.047), but without any interactions with parasite load (i.e., no differences in host tolerance) (Fig. 4). Host body size, which was larger among hosts from high-elevation ponds (Gaussian GLM; βhost_high = 0.776 ± 0.156, P < 0.0001), decreased most sharply for hosts with high loads of low-elevation parasites (LM; βload-by-parasite_low = −0.444 ± 0.200, P = 0.026), with no interactions between host source and load (Fig. 4).

Figure 4.

Figure 4.

Open in a new tab

Experimental partitioning of parasite population source and infection load on the processes of infection and pathology. (A) Effects of parasite population source (low or high elevation) on the number of encysting R. ondatrae metacercariae within P. regilla hosts as a function of parasite exposure dosage. (B) Differential effects of parasite load on host survival to 20 days as a function of parasite population source (low or high elevation). Host source is not depicted in the plots because it had no significant influence on resistance or tolerance. (C) Within-host distribution of R. ondatrae metacercariae among experimentally exposed hosts, illustrating the proportion of parasites found around the hind limb region as a function of parasite population source. Throughout, dashed peach lines reflect high-elevation parasite populations and teal solid lines reflects low-elevation populations.

Tadpoles exposed experimentally to R. ondatrae from low-elevation ponds also exhibited a greater fraction of encysted parasites around the hind limb region relative to those infected with parasites from high-elevation ponds (Binomial GLMM; βparasite_low = 2.93 ± 0.311, P < 0.0001). Overall, 92.2% of low-elevation parasites were observed around the hind limbs of their hosts, compared with only 35.8% of parasites from high elevation (Fig. 4). There was no effect of exposure dose, host source pond, or its interaction with parasite source, while host body size had a negative effect on the proportion of parasites near the hind limbs (βhost_size = −0.246 ± 0.106, P = 0.021).

Testing parasite source population effects in a second host species

Among larval toads (A. boreas) exposed to R. ondatrae, we observed a similarly strong difference in the fraction of parasites associated with the hind limbs as a function of parasite source. While 89.2% of low-elevation parasites were found around the hind limbs, this fraction dropped to 24.6% within hosts exposed to high-elevation parasites (Binomial GLMM; βparasite_low = −3.23 ± 0.524, P < 0.0001) (Fig. S2A). In contrast to P. regilla, A. boreas hosts exposed to low-elevation parasites supported higher average parasite loads compared with those exposed to the same dose of high-elevation cercariae (overdispersed Poisson GLMM; βparasite_low = 0.623 ± 0.154, P < 0.0001) (Fig. S2B). Toad body size had no effect on the distribution of metacercariae or on infection success, and no hosts died during the brief exposure window (36 hours).

Discussion

Virulence is central to research on disease ecology and evolution, but it is often defined inconsistently, difficult to measure empirically, and can be challenging to understand mechanistically. In part, this stems from the fact that virulence is classically considered a property of the parasite but is measured as a loss in fitness for the host. Yet growing evidence highlights the potential contributions of both host and parasite in determining virulence (Koella & Turner 2007; Medzhitov et al. 2012; Knutie et al. 2017), underscoring the need for empirical approaches that can effectively partition these components (e.g., Råberg et al. 2009; Råberg 2014). Here we combined large-scale field surveys with reciprocally crossed infection experiments to understand remarkable differences in virulence outcomes emerging from interactions between amphibian hosts and the trematode parasite, Ribeiroia ondatrae.

Based on surveys of 110 ponds across a 2000 m elevational gradient, frog populations at low elevation suffered significantly higher frequencies of parasite-induced limb malformations, even after controlling for parasite load. Reciprocal infection experiments that factorially crossed host and parasite sources across a dose–response regimen indicated that these observed differences in virulence were a property of the parasite, rather than the host or their interaction. Low-elevation parasites were more pathogenic, inhibiting host growth and survival, and also exhibited a narrower distribution within their frog hosts. While high-elevation parasites were relatively non-specific in their distribution, occurring at similar frequencies around the head and limbs of hosts, those from low elevation clustered strongly around the hind limbs, helping to explain differences in malformations observed in the field. This distributional pattern persisted across mountain ranges (Cascades and Sierras) and was evident in a second amphibian species (western toads) exposed experimentally, collectively reinforcing its generality.

The experimental framework used here offers insights into the two major processes underlying variation in virulence: infection success and host damage. Each process has constituent components associated with the host (resistance and tolerance) and with the parasite (infectivity and per-parasite pathogenicity) (Råberg 2014), yet we are aware of no studies that have simultaneously quantified each of these components. Using an experimental approach that crossed hosts and parasites from high and low elevations, we found that infection success varied with parasite source, but not with host source or their interaction, emphasizing the influence of variability in parasite infectivity rather than host resistance within this system. By also incorporating a range of parasite exposure dosages, we further tested for host and parasite contributions to host damage as a function of parasite load. Host survival and growth declined with load, as expected; however, this effect was greater for parasites from low-elevation sources, revealing differences in per-parasite pathogenicity. There was no such interaction between host population source and infection, indicating an absence of significant differences in host tolerance.

This approach also highlighted that the within-host distribution of parasites differed between parasite population sources, offering mechanistic insight into observed variation in malformation risk (virulence) in nature. When hosts were exposed to low-elevation R. ondatrae cercariae, 92% of the resulting metacercariae aggregated around the hind limbs, compared to only 36% for high-elevation parasites. This fraction was unaffected by host source, total parasite load, or the interaction between host and parasite source locations. The exact mechanisms determining cercariae encystment location are not yet known, and we highlight the opportunity to apply emerging genomic technologies (e.g., RADSeq) to this question and shed light on the genetic basis of virulence in this system (e.g., Davey & Blaxter 2010; DeCandia et al. 2018). After contacting an amphibian host, cercariae of R. ondatrae actively move across the surface of the host until finding a suitable encystment location, suggesting infection site selection represents a deliberate ‘choice’ (see also Leung et al. 2010; Sears et al. 2013).

Our findings build upon and extend previous research highlighting the importance of within-host parasite distribution in driving virulence. Mechanistically, pathogenic effects of trematodes in second intermediate hosts are often strongly tied to the number of infecting parasites (load-dependent pathology) and to the specific location wherein they infect the host (Poulin 2007). Encystment in specific tissues or organs may be more likely to induce pathologies, some of which have the potential to help transmit the parasite to a subsequent host (Moore 2002; Thomas et al. 2005; Leung et al. 2010). For parasites that depend on trophic transmission, infection-induced changes in host appearance, behavior, or morphology that increase the likelihood an infected intermediate host is consumed by an appropriate definitive host may be adaptively favored. For instance, encystment of Diplostomum spathaceum in the eyes of their freshwater fish intermediate hosts can cause cataracts that likely increase their vulnerability to predators (Seppälä et al. 2005). Similarly, cercariae of the trematode Curtuteria australis that infect the foot tip of cockles (Austrovenus stutchburyi) hinder their burrowing ability and expose them to greater predation risk by oystercatchers (Thomas and Poulin 1998), whereas no such effects are observed when parasites establish in the base of the foot. In some cases, within-host parasite aggregation may help to lessen pathology and avoid host mortality prior to transmission. Echinostome metacercariae occur disproportionately in the right kidneys of their amphibian hosts, which is hypothesized to limit severe damage in heavy infections (i.e., hosts can survive without one kidney but not without both) (Johnson et al. 2014).

What is especially interesting about R. ondatrae infection in amphibians is that the within-host distribution preference of invading parasites varies between populations, offering an explanation for observed variation in virulence outcomes in natural systems. Thus, changes in the proportion of parasites colonizing the hind limb region of hosts – through the resulting effects on host malformation risk – may reflect variation in optimal virulence investment. In a parallel example involving the trematode Microphallus papillorobustus, parasites can infect their amphipod intermediate hosts either around the abdomen, where they cause no apparent damage, or in the cerebroid ganglia, which is associated with altered host behaviors (“crazy gammarids”) that increase predation by bird definitive hosts (Thomas et al. 2000). Intriguingly, survival of metacercariae in the brain is lower than for those in the abdomen, suggesting that host manipulation may come at a cost to infection success. We observed a similar reduction in infection success for low-elevation R. ondatrae aggregated around host limbs, which exhibited a 17.5% infection success compared with 29.6% for parasites from high elevation. This apparent tradeoff between parasite infectivity and per-parasite pathogenicity may indicate that “choosiness” for encystment location comes at a cost to cercarial infection success, for example if greater searching time or energetic resources are needed to locate the hind limb region.

An intriguing question is why R. ondatrae induces higher virulence (more malformations) at low elevation. At least two classes of hypotheses can be advanced to account for this pattern. First, parasites could be genetically similar between low- and high-elevation areas, but differences in environmental conditions induce the development of more-virulent cercariae at lower elevations. Although our experimental approach standardized conditions between cercariae and tadpoles, there might nonetheless be differences in snail hosts between elevations or additional aspects of the environment, such as winter conditions, that affect pathogenicity. Second, parasites from the two regions could be genetically distinct from one another, such that the observed differences stem from evolutionary changes owing to either natural selection or genetic drift. From a selection standpoint, parasites will invest in host manipulation so long as the benefits to transmission outweigh the costs (Poulin 1994; Ebert & Herre 1996). For instance, if amphibian metapopulations are larger or more interconnected at low elevation (in contrast to those from more isolated mountain environments), high virulence may be less likely to feedback negatively on transmission (Gandon 2002; Boots et al. 2009). Even localized extirpations of amphibian hosts associated with infection may have few adverse consequences for transmission if hosts are readily available from nearby sources. While relatively little is known about the genetic structure of R. ondatrae populations, the low- and high-elevation populations studied here are separated by distances of ~250 km, which creates the potential for genetic isolation. In a previous study, we detected <0.5% sequence differences in CO1 between R. ondatrae populations from the Bay Area and Mt. Lassen (see Johnson et al. 2021), although next generation approaches might afford greater resolution and sensitivity.

From an applied perspective, these findings have relevance to amphibian conservation and disease management. Studies on the consequences of virulence often focus on changes in parasite load (e.g., Marcogliese 2008), yet our results illustrate that increases in per-parasite pathogenicity – independent of average load – can also lead to greater virulence. This suggests that the same disease can be problematic for one host population, but not another, despite similar parasite loads. (e.g., see Sears et al. 2013; Carnegie et al. 2021). For amphibians, ascertaining where and why R. ondatrae is virulent can inform conservation strategies. Infection by R. ondatrae can cause substantial mortality in amphibian populations that may exceed 90% in sensitive taxa (Wilber et al. 2020). In these instances, efforts may be needed to bolster population growth (to offset losses resulting from disease) or mitigate disease impacts. While disease mitigation efforts frequently look to modify the host, by enhancing host resistance or reducing density, these results emphasize the parallel importance of parasite source population in driving disease outcomes. This variation may even be climate-driven, in that the higher virulence parasites we observed occur in lower and warmer environments, consistent with other disease systems in which temperature has been positively linked to virulence (Marcogliese 2008; Pulkkinen et al. 2010; Kirk et al. 2018; Hector et al. 2023). Looking forward, we underscore the value of studies that identify the mechanistic basis of differential virulence and evaluate how virulence outcomes change with alterations to hosts, parasites, or their environments (e.g., Decker et al. 2018).

Supplementary Material

Supinfo1
Sfig2
SFig1

Acknowledgments

We gratefully acknowledge the many individuals who helped make this research possible. For assistance with field work and collecting animals for experiments, we thank: J. Bowerman, S. Kupferberg, K. Lunde, K. McCaffrey, W. Moss, S. Paull, D. Rose, I. Chellman, L. Earley, M. Magnuson, T. Rickman, J. Beasley, E. White, and A. Kalonia. For their help in quantifying parasite infection and conducting laboratory experiments, we recognize: E. Esfahani, K. Leslie, T. Riepe, J. Curtis, E. Hannon, C. Hassan, N. Pelton, R. Seyler, and E. Ursich. Property access to field sites was generously provided by: Lassen Volcanic National Park, Lassen National Forest, East Bay Regional Parks District, East Bay Municipal Utility District, Santa Clara County Parks, and Blue Oaks Ranch Reserve. This research was funded through: the United States Department of Agriculture Forest Services (17-CS-11050600-015), the National Science Foundation (DEB-0841758, DEB-1149308, DEB-1754171, IOS-1754886), the National Institutes of Health (R01GM109499, R01GM135935), a Section 6 Grant from the California Department of Fish and Wildlife and US Fish and Wildlife Service (P188010), the David and Lucile Packard Foundation, and the University of Colorado Undergraduate Research Opportunity Program.

Data accessibility statement:

The data used in this study are available through Figshare at doi.org/10.6084/m9.figshare.22782212.

References

  1. Boots M, Best A, Miller MR & White A (2009). The role of ecological feedbacks in the evolution of host defence: what does theory tell us? Philos. Trans. R. Soc. London, Ser. B, 364, 27–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Brooks ME, Kristensen K, Van Benthem KJ, Magnusson A, Berg CW, Nielsen A et al. (2017). glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. R Journal, 9, 378–400. [Google Scholar]
  3. Carnegie RB, Ford SE, Crockett RK, Kingsley-Smith PR, Bienlien LM, Safi LS et al. (2021). A rapid phenotype change in the pathogen Perkinsus marinus was associated with a historically significant marine disease emergence in the eastern oyster. Sci. Rep, 11, 12872. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Davey JW & Blaxter ML (2010). RADSeq: next-generation population genetics. Brief Funct. Genomics, 9, 416–423. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. DeCandia AL, Dobson AP & vonHoldt BM (2018). Toward an integrative molecular approach to wildlife disease. Conserv. Bio, 32, 798–807. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Decker LE, de Roode JC & Hunter MD (2018). Elevated atmospheric concentrations of carbon dioxide reduce monarch tolerance and increase parasite virulence by altering the medicinal properties of milkweeds. Ecol. Lett, 21, 1353–1363. [DOI] [PubMed] [Google Scholar]
  7. Ebert D & Herre EA (1996). The evolution of parasitic diseases. Parasitol. Today, 12, 96–101. [DOI] [PubMed] [Google Scholar]
  8. Gandon S, Agnew P & Michalakis Y (2002). Coevolution between parasite virulence and host life-history traits. Am. Nat, 160, 374–388. [DOI] [PubMed] [Google Scholar]
  9. Gibson AK & Amoroso CR (2022). Evolution and ecology of parasite avoidance. Annu. Rev. Ecol., Evol., and Syst, 53, 47–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. González R, Butković A, Escaray FJ, Martínez-Latorre J, Melero Í, Pérez-Parets E et al. (2021). Plant virus evolution under strong drought conditions results in a transition from parasitism to mutualism. PNAS, 118, e2020990118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Goodman BA & Johnson PTJ (2011). Disease and the extended phenotype: parasites control host performance and survival through induced changes in body plan. PLoS One, 6, e20193. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Gosner KL (1960). A simplified table for staging anuran embryos and larvae with notes on identification. Herpetologica, 16, 183–190. [Google Scholar]
  13. Greischar MA & Koskella B (2007). A synthesis of experimental work on parasite local adaptation. Ecol. Lett, 10, 418–434. [DOI] [PubMed] [Google Scholar]
  14. Hector T, Gehman AL & King K (2023). Infection burdens and virulence under heat stress: ecological and evolutionary considerations. Philos. Trans. R. Soc. London, Ser. B, 378, 20220018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hollister J. (2021). Accessing elevation data in R with the elevatr package. Available at: https://cran.rproject.org/web/packages/elevatr/vignettes/introduction_to_elevatr.html#Introduction_to_elevatr. Last accessed 2/17/2023. [Google Scholar]
  16. Holmes JC (1973). Site selection by parasitic helminths: interspecific interactions, site segregation, and their importance to the development of helminth communities. Can. J. Zool, 51, 333–347. [DOI] [PubMed] [Google Scholar]
  17. Johnson PTJ, Calhoun DM, Moss WE, McDevitt-Galles T, Riepe TB, Hallas JM et al. (2021). The cost of travel: how dispersal ability limits local adaptation in host–parasite interactions. J. Evol. Biol, 34, 512–524. [DOI] [PubMed] [Google Scholar]
  18. Johnson PTJ, Kellermanns E & Bowerman J (2011). Critical windows of disease risk: amphibian pathology driven by developmental changes in host resistance and tolerance. Funct. Ecol, 25, 726–734. [Google Scholar]
  19. Johnson PTJ, Koprivnikar J, Orlofske SA, Melbourne BA & Lafonte BE (2014). Making the right choice: testing the drivers of asymmetric infections within hosts and their consequences for pathology. Oikos, 123, 875–885. [Google Scholar]
  20. Johnson PTJ, Lunde KB, Thurman EM, Ritchie EG, Wray SN, Sutherland DR et al. (2002). Parasite (Ribeiroia ondatrae) infection linked to amphibian malformations in the western United States. Ecol. Monogr, 72, 151–168. [Google Scholar]
  21. Johnson PTJ, Rohr JR, Hoverman JT, Kellermanns E, Bowerman J & Lunde KB (2012). Living fast and dying of infection: host life history drives interspecific variation in infection and disease risk. Ecol. Lett 15: 235–242 [DOI] [PubMed] [Google Scholar]
  22. Johnson PTJ, Preston DL, Hoverman JT & Richgels KLD (2013). Biodiversity reduces disease through predictable changes in host community competence. Nature, 494, 230–234. [DOI] [PubMed] [Google Scholar]
  23. Johnson PTJ, Sutherland DR, Kinsella JM & Lunde KB (2004). Review of the trematode genus Ribeiroia (Psilostomidae): ecology, life history, and pathogenesis with special emphasis on the amphibian malformation problem. Adv. Parasitol, 57, 191–253. [DOI] [PubMed] [Google Scholar]
  24. Kirk D, Jones N, Peacock S, Phillips J, Molnár PK, Krkošek M et al. (2018). Empirical evidence that metabolic theory describes the temperature dependency of within-host parasite dynamics. PLoS Biol., 16, e2004608. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Klemme I, Hyvärinen P & Karvonen A (2020). Negative associations between parasite avoidance, resistance and tolerance predict host health in salmonid fish populations. Proc. R. Soc. B, 287, 20200388. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Knutie SA, Wilkinson CL, Wu QC, Ortega CN & Rohr JR (2017). Host resistance and tolerance of parasitic gut worms depend on resource availability. Oecologia, 183, 1031–1040. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Koella JC & Turner P (2007). Evolution of parasites. In: Evolution medicine and public health (ed. Stearns Stephen C., JCK). Oxford University Press Oxford, pp. 229–238. [Google Scholar]
  28. Little TJ, Shuker DM, Colegrave N, Day T & Graham AL (2010). The coevolution of virulence: tolerance in perspective. PLoS Pathog., 6, e1001006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Lunde KB & Johnson PTJ (2012). A practical guide for the study of amphibian malformations and their causes. J. Herp, 46, 429–441. [Google Scholar]
  30. Leung TL, Keeney DB & Poulin R (2010). Genetics, intensity-dependence, and host manipulation in the trematode Curtuteria australis: following the strategies of others? Oikos, 119, 393–400. [Google Scholar]
  31. Marcogliese DJ (2008). The impact of climate change on the parasites and infectious diseases of aquatic animals. Rev. Sci. Tech, 27, 467–484. [PubMed] [Google Scholar]
  32. McDevitt-Galles T, Moss WE, Calhoun DM & Johnson PTJ (2020). Phenological synchrony shapes pathology in host–parasite systems. Proc. Royal Soc. B, 287, 20192597. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Medzhitov R, Schneider DS & Soares MP (2012). Disease tolerance as a defense strategy. Science, 335, 936–941. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Moore J. (2002). Parasites and the behavior of animals. Oxford University Press, Oxford. [Google Scholar]
  35. Poulin R. (2007). Evolutionary Ecology of Parasites. Princeton University Press, Princeton [Google Scholar]
  36. Poulin R, Brodeur J & Moore J (1994). Parasite manipulation of host behaviour: should hosts always lose? Oikos, 70, 479–484. [Google Scholar]
  37. Pulkkinen K, Suomalainen LR, Read AF, Ebert D, Rintamäki P & Valtonen ET (2010). Intensive fish farming and the evolution of pathogen virulence: the case of columnaris disease in Finland. Proc. Royal Soc. B, 277, 593–600. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. R Core Team (2022). R: A language and environment for statistical computing. Available at: https://www.R-project.org/ Last accessed 2/27/2023.
  39. Råberg L. (2014). How to live with the enemy: understanding tolerance to parasites. PLoS Biol., 12, e1001989. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Råberg L, Graham AL & Read AF (2009). Decomposing health: tolerance and resistance to parasites in animals. Philos. Trans. R. Soc. London, Ser. B, 364, 37–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Råberg L, Sim D & Read AF (2007). Disentangling genetic variation for resistance and tolerance to infectious diseases in animals. Science, 318, 812–814. [DOI] [PubMed] [Google Scholar]
  42. Rohr JR, Raffel TR & Hall CA (2010). Developmental variation in resistance and tolerance in a multi-host–parasite system. Funct. Ecol, 24, 1110–1121. [Google Scholar]
  43. Schmid-Hempel P. (2009). Immune defence, parasite evasion strategies and their relevance for ‘macroscopic phenomena’such as virulence. Philos. Trans. R. Soc. London, Ser. B, 364, 85–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Sears BF, Snyder PW & Rohr JR (2013). Infection deflection: hosts control parasite location with behaviour to improve tolerance. Proc. Royal Soc. B, 280, 20130759. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Seppälä O, Karvonen A & Valtonen ET (2005). Manipulation of fish host by eye flukes in relation to cataract formation and parasite infectivity. Anim. Behav, 70, 889–894. [Google Scholar]
  46. Stewart Merrill TE, Calhoun DM & Johnson PTJ (2022). Beyond single host, single parasite interactions: Quantifying competence for complete multi-host, multi-parasite communities. Funct. Ecol, 36, 1845–1857. [Google Scholar]
  47. Stewart Merrill TE & Johnson PTJ (2020). Towards a mechanistic understanding of competence: a missing link in diversity–disease research. Parasitology, 147, 1159–1170. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Tadiri CP, Fussmann GF & Scott ME (2021). Parasite spread in experimental metapopulations: resistance, tolerance and host competence. Oikos, 130, 1562–1571. [Google Scholar]
  49. Thomas F, Adamo S & Moore J (2005). Parasitic manipulation: where are we and where should we go? Behav. Processes, 68, 185–199. [DOI] [PubMed] [Google Scholar]
  50. Thomas F, Guldner E & Renaud F (2000). Differential parasite (Trematoda) encapsulation in Gammarus aequicauda (Amphipoda). J. Parasitol, 86, 650–654. [DOI] [PubMed] [Google Scholar]
  51. Thomas F & Poulin R (1998). Manipulation of a mollusc by a trophically transmitted parasite: convergent evolution or phylogenetic inheritance? Parasitology, 116, 431–436. [DOI] [PubMed] [Google Scholar]
  52. Thomas MB & Blanford S (2003). Thermal biology in insect-parasite interactions. Trends. Ecol. Evol, 18, 344–350. [Google Scholar]
  53. Wilber MQ, Briggs CJ & Johnson PTJ (2020). Disease's hidden death toll: Using parasite aggregation patterns to quantify landscape-level host mortality in a wildlife system. J Anim Ecol, 89, 2876–2887. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supinfo1
NIHMS1951102-supplement-Supinfo1.docx (13.4KB, docx)
Sfig2
NIHMS1951102-supplement-Sfig2.docx (61.6KB, docx)
SFig1
NIHMS1951102-supplement-SFig1.docx (105.2KB, docx)

Data Availability Statement

The data used in this study are available through Figshare at doi.org/10.6084/m9.figshare.22782212.

RESOURCES