Introduction
Aquatic environments often contain a diverse community of macroparasite species that make up a significant portion of free-living biomass, play an important role in food webs and serve as a source of infectious disease (Hechinger and Lafferty, Reference Hechinger and Lafferty2005; Lafferty et al., Reference Lafferty, Allesina, Arim, Briggs, De Leo, Dobson, Dunne, Johnson, Kuris, Marcogliese, Martinez, Memmott, Marquet, McLaughlin, Mordecai, Pascual, Poulin and Thieltges2008; Johnson and Hoverman, Reference Johnson and Hoverman2012; Preston et al., Reference Preston, Orlofske, Lambden and Johnson2013). Indeed, this parasite biodiversity has led to a recognition of the importance of research focusing on how multi-parasite interactions influence infection outcomes within a host (Johnson and Hoverman, Reference Johnson and Hoverman2012; Hoverman et al., Reference Hoverman, Hoye and Johnson2013; Wuerthner et al., Reference Wuerthner, Hua and Hoverman2017). This research has shown that when a host is challenged by two parasites simultaneously, the coinfection can increase, decrease or have no effect on the competitive ability of each parasite, depending on factors such as competition for resources and cross-reactive immunity (Pedersen and Fenton, Reference Pedersen and Fenton2007). Moreover, hosts are often challenged by different parasite species sequentially due to spatial, diel and seasonal variations in prevalence and emergence (Fingerut et al., Reference Fingerut, Zimmer and Zimmer2003, Studer and Poulin, Reference Studer and Poulin2012, Marino et al., Reference Marino, Holland and Werner2017). Thus, priority effects driven by both exposure timing and order can also strongly influence infection dynamics and result in different outcomes compared to simultaneous infection. For example, Hoverman et al. (Reference Hoverman, Hoye and Johnson2013) found that infection of larval amphibians with echinostome trematodes prior to exposure to the trematode species Ribeiroia ondatrae reduced R. ondatrae infection success. Because multiple parasite species commonly co-occur and the prevalence of different parasite species is variable through space and time, studies that elucidate the consequences of coinfection on infection outcomes are important for a broader understanding of disease dynamics in natural populations (Telfer et al., Reference Telfer, Birtles, Bennett, Lambin, Paterson and Begon2008, Reference Telfer, Lambin, Birtles, Beldomenico, Burthe, Paterson and Begon2010; Ezenwa et al., Reference Ezenwa, Etienne, Luikart, Beja-Pereira and Jolles2010; Knowles, Reference Knowles2011).
Echinostomes (Echinostomatidae) are a diverse and common group of trematode parasites, with molecular evidence suggesting that over ten echinostome species exist in the United States (Detwiler et al., Reference Detwiler, Bos and Minchella2010). Echinostomes are frequently used as a model parasite in ecological studies, particularly in amphibian systems due to their ability to cause intensity-dependent pathology and disease (Fried et al., Reference Fried, Pane and Reddy1997; Holland et al., Reference Holland, Skelly, Kashgarian, Bolden, Harrison and Cappello2007). Much of the research on echinostome-driven pathology in amphibians has focused on one species, Echinostoma trivolvis (Szuroczki and Richardson, Reference Szuroczki and Richardson2009). However, despite slight morphological variations between some echinostome species, echinostomes can be difficult to differentiate without molecular tools (Kostadinova et al., Reference Kostadinova, Herniou, Barrett and Littlewood2003; Detwiler et al., Reference Detwiler, Bos and Minchella2010). Because of this, many experimental studies use field-collected echinostomes with identification limited to family, and some field surveys have likely underestimated the true macroparasite biodiversity in a system (Koprivnikar et al., Reference Koprivnikar, Forbes and Baker2007; Hoverman et al., Reference Hoverman, Mihaljevic, Richgels, Kerby and Johnson2012, Reference Hoverman, Hoye and Johnson2013; Buss and Hua, Reference Buss and Hua2018). The frequent co-occurrence of multiple echinostome species suggests that some experimental studies may have used two or more species (Detwiler et al., Reference Detwiler, Bos and Minchella2010, Reference Detwiler, Zajac, Minchella and Belden2012), with the potential for interactions within the host that may affect infection success and pathology. However, the interactions between different echinostome species in second intermediate hosts have not been well explored (but see Leung and Poulin, Reference Leung and Poulin2011).
Amphibians present an excellent model system for assessing the influence of timing and order of exposure to multiple echinostome species on infection outcomes. Amphibians are a second intermediate host for a variety of trematode species and frequently co-occur with echinostomes (Hoverman et al., Reference Hoverman, Mihaljevic, Richgels, Kerby and Johnson2012). Echinostomes and other trematodes typically use freshwater snails as a first intermediate host. The rediae in the first intermediate host then release free-swimming cercariae into the water column, which enter tadpoles via the cloacal opening and encyst in the kidney as metacercariae (Szuroczki and Richardson, Reference Szuroczki and Richardson2009). Echinostome cercariae typically encyst preferentially in the right kidney, which has been viewed as an evolutionary adaptation to reduce parasitic impairment of both organ structures and thus lessen the impact of infection on host and parasite fitness (Thiemann and Wassersug, Reference Thiemann and Wassersug2000; Johnson et al., Reference Johnson, Koprivnikar, Orlofske, Melbourne and Lafonte2014). Because there is a considerable number of field and experimental studies of the echinostome-amphibian system (Johnson and McKenzie, Reference Johnson, McKenzie, Toledo and Fried2008), it is of particular interest to explore how different echinostome species vary in infection success and how coinfection with two or more co-occurring echinostome species may influence infection outcomes.
Here, we used observational field data of echinostome prevalence across three months in seven ponds to inform laboratory experiments that explore how coinfection with two commonly co-occurring echinostome species influences infection outcomes in larval leopard frogs (Rana pipiens). Specifically, we examined how the timing of host exposure to two common echinostomes (E. trivolvis and Echinoparyphium lineage 3), presented separately and in combination, influences patterns of infection. When using single parasite exposures, we predicted that direct interspecific interactions would not strongly influence infection loads and that infection loads would be similar when the parasite species are present separately and in combination. When using two sequential parasite exposures, we predicted tadpoles exposed to the two parasite species sequentially would have higher infection loads than tadpoles exposed to the same parasite species sequentially, due to negative intraspecific interactions that would prime the immune system (Cox, Reference Cox2001).
Material and methods
Field surveys
As part of a broader project to collect Helisoma trivolvis snails infected with E. trivolvis, we surveyed seven ponds in northwestern Pennsylvania from May to July 2018. We surveyed each pond once a month and collected H. trivolvis, which is a first intermediate host for echinostomes. We collected snails using unstandardized dipnet sweeps through submerged vegetation with the goal of collecting a minimum of 50 snails per site, keeping only snails >5 mm; snails below this threshold tend to be immature and rarely support infections (Richgels et al., Reference Richgels, Hoverman and Johnson2013). The number of collected snails varied by site and month [Month: mean (range); May: 143.7 (32–330); June: 235.7 (125–300); July: 175 (100–200)].
To screen collected snails for echinostome infection, we isolated individual snails in 50 mL tubes filled with 35 mL of UV-irradiated well water and placed them 10 cm below a light source for 1 h to induce cercarial shedding (Szuroczki and Richardson, Reference Szuroczki and Richardson2009). We identified echinostome-infected snails by placing cercariae on slides under a compound scope following Schell (Reference Schell1985). We then isolated a single cercaria from each echinostome-infected snail and preserved it in 95% ethanol. Although it is possible that some snails were infected with multiple echinostome species, the overall prevalence of double infections is typically low (~2.5%; Sousa, Reference Sousa1993; Lafferty et al., Reference Lafferty, Sammond and Kuris1994). Moreover, experimental work has shown that double infection with echinostomatids is particularly unstable; one species is typically eliminated by the other (Dönges, Reference Dönges1972). This makes it possible, but unlikely, that any individual snail was double infected and shedding cercariae from two echinostome species.
Trematode identification
We extracted the genomic DNA from each trematode sample using a Qiagen DNeasy extraction kit. The internal transcribed spacer region (ITS-1) of ribosomal DNA was amplified by polymerase chain reaction (PCR), using forward (BD1, 5′ – GTC GTA ACA AGG TTT CCG TA – 3′, Bowles and McManus, Reference Bowles and McManus1993) and reverse (4S, 5′ – TCT AGA TGC GTT CGA A(G/A)T GTC GAT G – 3′, Bowles and McManus Reference Bowles and McManus1993) primers. The cycling parameters for the PCR were 95 °C for 2 min followed by 35 cycles of 94°C for 1 min, 50 °C for 45 s and 72 °C for 1 min, finishing with an elongation at 72 °C for 7 min. We then cleaned the PCR products with QIAquick PCR Purification Kit (Qiagen). Sequencing was conducted by the Purdue Genomics Core Facility in the forward and reverse directions using a BigDye terminator kit (Applied Biosystems) and an ABI 3730XL sequencer.
A total of 89 ITS1 sequences were generated. We condensed these 89 sequences into 14 unique haplotypes using DnasSP v6 prior to the phylogenetic analysis (Rozas et al., Reference Rozas, Ferrer-Mata, Sanchez-DelBarrio, Guirao-Rico, Librado, Ramos-Onsins and Sanchez-Gracia2017). We aligned one sample representing each of the 14 haplotypes with 23 previously published sequences of echinostomatids (Detwiler et al., Reference Detwiler, Bos and Minchella2010) from GenBank automatically using the MUSCLE alignment in the program MEGA version X (Kumar et al., Reference Kumar, Stecher, Li, Knyaz and Tamura2018) and then rechecked manually by eye. Phylogenetic analysis was conducted with a Bayesian inference approach (MrBayes 3.2.7a; Ronquist et al., Reference Ronquist, Teslenko, Van Der Mark, Ayres, Darling, Höhna, Larget, Liu, Suchard and Huelsenbeck2012). We used the General Time Reversible plus Invariant sites plus Gamma distributed model of nucleotide substitution for our analysis. Two simultaneous Bayesian runs were conducted [with the default Markov chain Monte Carlo (MCMC) settings], and run for a total of 5.0 × 106 generations per run. We sampled trees and parameters every 100 generations and the first 25% of each run was discarded as burn-in (Hua et al., Reference Hua, Buss, Kim, Orlofske and Hoverman2016). We rooted the tree with three outgroup taxa. We then used this tree to identify our novel sequences based on Bayesian support values.
Experimental animal collection and husbandry
Based on our field surveys, the two most abundant echinostomatids at our field sites were Echinostoma trivolvis (hereafter Echinostoma) and Echinoparyphium lineage 3 (hereafter Echinoparyphium). We maintained a subset of the ten snails each infected with Echinostoma (site name – number of snails; LOG – 3; MAP – 3; CRK – 2; RMD – 2) and Echinoparyphium (site name – number of snails; LOG – 2; MAP – 3; CRK – 3; RMD – 1; JER – 1) in the lab for the experiments. Snails were housed individually in 1 L cups filled with 0.8 L of UV-irradiated well water and held at 7 °C to reduce shedding prior to the start of experiments. One day prior to the start of the experiments, snails were slowly acclimated to 23 °C. Snails were fed a mixture of rabbit chow and spirulina powder ad libitum.
We collected ten leopard frog (R. pipiens) egg masses from a pond at the Purdue Wildlife Area (PWA) in West Lafayette, IN on 12 April 2018. We distributed the egg masses into 180 L outdoor culturing pools that had been filled with aged well water and covered with 70% shade cloth to prevent predator oviposition. After hatching, tadpoles were fed rabbit chow (Purina) ad libitum until the start of experiments. Tadpole health was checked daily. Two days prior to the start of experiments, we brought 150 haphazardly chosen leopard frog tadpoles [snout-vent length (SVL) = 10.9 ± 1.1 mm s.d., stage = 27.5 ± 0.7 s.d.; Gosner, Reference Gosner1960] into the lab to acclimate to indoor conditions (23 °C, 12:12 light cycle).
Experiment 1: simultaneous exposures
The first experiment examined whether infection loads differed between exposure to each echinostome species separately or combined. A single tadpole was assigned to each experimental unit. Experimental units were completely randomized across a shelving unit and assigned to one of three treatments: (1) exposure to 50 Echinostoma cercariae; (2) exposure to 50 Echinoparyphium cercariae; or (3) exposure to 25 Echinostoma cercariae and 25 Echinoparyphium cercariae simultaneously (N = 10 tadpoles per treatment). Initial parasite exposures took place in 130 mL cups filled with 75 mL UV-irradiated well water. After 12 h, the tadpoles were transferred to a 1 L cup filled with 800 mL of UV-irradiated well water, as cercariae efficacy is highest during the first 8 h after shedding and declines with time (Rohr et al., Reference Rohr, Raffel, Sessions and Hudson2008).
To obtain cercariae to add to the experimental units, we individually shed the echinostome-infected snails for 1 h, as described previously, and then homogenized the cercariae by species in 1 L cups. We collected and counted free-swimming cercariae with a glass pipette under a dissecting scope and transferred the cercariae directly into each experimental unit. While dosing experimental units, we renewed the source parasites in 1 L cups every hour by discarding the unused cercariae and adding newly shed cercariae so that all cercariae entered experimental units within 2 h of emergence from the snail, which is when they are most infective.
We checked for tadpole mortality daily for 5 days and fed each tadpole rabbit chow ad libitum during the experiment. We ended the experiment after 5 days, as this allows echinostome cercariae to successfully encyst in the kidneys and minimizes time for tadpoles to clear infection (Hoverman et al., Reference Hoverman, Hoye and Johnson2013). We euthanized (MS-222 overdose) and individually preserved tadpoles in 70% ethanol. Prior to necropsy, we weighed, staged and measured SVL and the total length of each tadpole. To quantify infection load, we first dissected the left and right kidney structures (primary kidney, nephric duct, pronephros; hereafter referred to as left and right kidneys) of each tadpole, placed them between two microscope slides, and counted the total number of metacercariae on each side under a compound microscope (Schotthoefer et al., Reference Schotthoefer, Cole and Beasley2003). We also examined the rest of the body for metacercariae, but all cysts were found in the kidneys.
Experiment 2: sequential exposures
The second experiment examined how the timing and sequence of exposure of tadpoles to Echinostoma and Echinoparyphium influence total echinostome infection loads. Each experimental unit was randomly assigned to a factorial treatment combination of parasite exposure on day 0 (25 Echinostoma, 25 Echinoparyphium or no parasites) and parasite exposure on day 3 (25 Echinostoma, 25 Echinoparyphium or no parasites). The nine treatments are summarized in Table 1. The experimental setup and parasite exposure methods were identical to experiment 1. Tadpoles were exposed to parasites on two dates, day 0 and day 3. On each day, we added either 25 cercariae of the appropriate species or no cercariae to each experimental unit. We checked for tadpole mortality daily and fed each tadpole rabbit chow ad libitum during the experiment. The experiment was ended on day 5, 2 days after the second exposure. We euthanized (MS-222 overdose) and individually preserved tadpoles in 70% ethanol. Tadpoles were necropsied as described in experiment 1. Because experiment 1 and experiment 2 occurred simultaneously in the same randomized experimental array, we draw comparisons between the two experiments.
Table 1. Summary of the 12 treatments used in experiments 1 and 2
In experiment 1, Rana pipiens tadpoles experienced a single exposure to 50 Echinostoma trivolvis, 50 Echinoparyphium lineage 3 or 25 cercariae of each species on day 0. In experiment 2, R. pipiens tadpoles were exposed to a factorial combination of 25 E. trivolvis, 25 Echinoparyphium lin. 3 or no parasites.
Statistical analyses
Because our field surveys were unstandardized and limited in scope, we restricted our analysis of the field data to summary statistics and qualitative descriptions to inform our experimental design.
Our main response variable for the experimental studies was overall infection load (count of metacercariae recovered). Although there are slight size variations between the two echinostome species (Fried et al., Reference Fried, Frazer and Kanev1998), this is not a reliable metric to differentiate a large quantity of metacercarial cysts, and molecular identification of so many parasites is not feasible. We used a negative binomial generalized linear model (nbGLM) for these analyses as our count data were overdispersed. To examine whether infection loads differed among single exposures to Echinostoma cercariae, Echinoparyphium cercariae or a combination of both parasites in experiment 1, we conducted an nbGLM with treatment as the main predictor and final SVL as a covariate, as tadpole size can influence susceptibility (Marino et al., Reference Marino, Holland and Werner2017). To examine how the timing and sequence of exposure of tadpoles to Echinostoma and Echinoparyphium cercariae affected total echinostome infection loads in experiment 2, we conducted an nbGLM with treatment as the main predictor and final SVL as a covariate. We also conducted a second nbGLM with day 0 treatment and day 3 treatment as the main predictors and final SVL as a covariate to determine if differences in infection loads differed based on the species used for the initial or secondary exposure.
To better determine if intraspecific or interspecific interactions influenced total echinostome infection loads in experiments 1 and 2, we compared the observed metacercarial load of each treatment to an expected value (Hoverman et al., Reference Hoverman, Hoye and Johnson2013). The expected parasite infection load was calculated by adding together the average number of parasites recovered from treatments exposed to only 25 cercariae on either day 0 or day 3 (Table 1); this value represents the infection load that would be expected if infection was additive (i.e. no antagonistic or synergistic interactions). This value was then compared to the observed parasite infection load for the treatments exposed to 50 total cercariae (Table 1). For example, to determine if interspecific interactions influenced infection load in the treatment exposed to 25 Echinostoma + 25 Echinoparyphium simultaneously on day 0, we first added the average number of parasites recovered from the treatment exposed to 25 Echinostoma on day 0 (mean = 9.1 metacercariae) with the average number of parasites recovered from the treatment exposed to 25 Echinoparyphium on day 0 (mean = 8.9 metacercariae). We then used a single sample t-test to determine whether the observed infection load differed from the expected infection load for each of the treatments receiving a total of 50 cercariae. We also conducted a series of repeated G-tests of goodness-of-fit to compare the observed distribution of metacercariae recovered from each kidney (left and right) with those expected assuming completely random, equal distribution on the left and right side.
We performed all statistical analyses using R version 3.5.1 (R Core Team, 2018). We determined that our metacercarial cyst load count data for was overdispersed and that using a Poisson distribution was inappropriate using the dispersiontest() function in the ‘AER’ package (dispersion = 2.45, Z = 5.24, P < 0.001; Kleiber and Zeileis, Reference Kleiber and Zeileis2008). Negative binomial GLMs were conducted using the glm.nb() function in the ‘MASS’ package (Venables and Ripley, Reference Venables and Ripley2002). We used the Anova() function in the ‘car’ package to estimate P values (Fox and Weisberg, Reference Fox and Weisberg2011). Estimated marginal means were calculated with the ‘emmeans’ package (Lenth et al., Reference Lenth, Singmann, Love, Buerkner and Herve2019), and Tukey post hoc tests were used to determine where significant differences among the treatments occurred with the cld() function in the ‘multcomp’ package (Hothorn et al., Reference Hothorn, Bretz and Westfall2008). We used the t.test() function to compare observed and expected metacercarial loads. Figures were made using ‘ggplot2’ (Wickham, Reference Wickham2016). We did not log-transform SVL, as it was normally distributed, and transformation did not improve normality. We excluded four individual tadpoles (but no more than one in any treatment) from analyses as their kidneys were degraded prior to cyst counting.
Research ethic approval for infection of tadpoles
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional guides on the care and use of laboratory animals. The authors have involved the minimum number of animals to produce statistically reproducible results. All methods for the infection of tadpoles were approved by the Purdue University IACUC (protocol 1701001530). Animals were collected under the Indiana Department of Natural Resources permit 18-066.
Results
Echinostome identification and prevalence in natural ponds
Using the sequenced ITS region from individual cercariae, each obtained from a different H. trivolvis snail host, we identified 89 echinostome samples consisting of 14 distinct haplotypes in the field survey. By comparing these 14 haplotypes to previously published sequences of echinostomatids (Detwiler et al., Reference Detwiler, Bos and Minchella2010; Fig. 1), we identified our samples as belonging to three species: E. trivolvis (haplotypes 10–13, N = 37), Echinostoma revolutum (haplotype 14, N = 1) and Echinoparyphium lineage 3 (haplotypes 1–9, N = 51) (Table 2). E. trivolvis and Echinoparyphium lineage 3 were identified from snails at all seven sites (Fig. 2). Both species were identified at 13 out of 21 samplings and both species co-occurred in 7/21 samplings at five different sites. The single E. revolutum sample was identified from a June sampling at one pond. Prevalence in snail hosts by month and site ranged from 0 to 6% for E. trivolvis and 0–12.5% for Echinoparyphium lineage 3. Our field surveys provided substantial evidence that E. trivolvis and Echinoparyphium lineage 3 commonly co-occur in our study region.
Fig. 1. Phylogenetic estimate of relationships within Echinostomatidae based on the ITS1 gene inferred from Bayesian support values and rooted with three outgroup species. Support values are shown near the nodes. Nodes supported by ⩾95% posterior probability are considered highly supported. The circles denote the species or lineages which were detected in our sampling.
Fig. 2. The prevalence of Echinoparyphium lineage 3 and Echinostoma trivolvis in seven ponds in northwest Pennsylvania over a three-month survey in 2018. Each bar depicts the proportion of infected Helisoma trivolvis snails, separated by trematode species.
Table 2. Haplotype frequencies of the 89 Echinostomatidae ITS1 sequences obtained from cercariae for this study.
Experiment 1: simultaneous exposures
For tadpoles that experienced a single exposure to separate or combined echinostomatid species, we found that infection loads significantly differed between the three treatments (χ 2 = 6.4, d.f. = 2, P = 0.041; Fig. 3A). The mean metacercarial load of individuals exposed to a mixed cohort of Echinostoma and Echinoparyphium was 58.2% higher than the mean metacercarial load of individuals exposed to Echinoparyphium alone (Z = 2.353, P = 0.048). Although individuals exposed to only Echinostoma tended to have higher infection loads than individuals exposed only to Echinoparyphium, this difference was not statistically significant (Z = −1.969, P = 0.120). Mean metacercarial load of individuals exposed to a mixed cohort of Echinostoma and Echinoparyphium did not differ from the mean metacercarial load of individuals exposed to only Echinostoma (Z = −0.313, P = 0.947). We did not find that infection loads in any treatment differed significantly from the expected value (t ⩽ 2.0, d.f. = 9, P ⩾ 0.076; Fig. 3A). Only one individual died during this experiment, so we did not assess the effects of treatment on mortality.
Fig. 3. Parasite infection loads (estimated marginal mean for number of metacercariae recovered ± s.e.) in Rana pipiens tadpoles after different combinations of exposure to two echinostome species at one or two time points. For each treatment, the day 0 exposure is shown above the day 3 exposure and separated by ‘—’. Panel A shows the treatments receiving 50 cercariae at once on day 0; Panel B shows the treatments receiving 25 cercariae on day 0 and 25 cercariae on day 3; Panel C shows the treatments receiving 25 cercariae on either day 0 or day 3. Within each panel, treatments sharing lower case letters are not significantly different from each other (P > 0.05). For each treatment in panel A and B, the expected total infection load calculated from mean infection loads in panel C is shown as an open circle. An asterisk denotes a significant difference between the expected and observed value (P < 0.05). Ep3 represents Echinoparyphium lineage 3 and Etr represents Echinostoma trivolvis.
Experiment 2: sequential exposures
For the hosts in the four treatments that experienced two exposures to either Echinostoma or Echinoparyphium on day 0 and day 3 (D–G, Table 2), we found evidence that mean metacercarial load was significantly different between treatments (χ 2 = 12.8, d.f. = 1, P = 0.005; Fig. 3B). Mean metacercarial load of individuals exposed to Echinoparyphium on day 0 and Echinostoma on day 3 was ~36.8% lower than either treatment with exposure to Echinostoma on day 0 (Z ⩾ 3.029, P ⩽ 0.013). The mean metacercarial load of individuals exposed to Echinoparyphium on day 0 and Echinoparyphium on day 3 did not significantly differ from the other three treatments (Fig. 3B). We also found evidence that mean metacercarial load was significantly influenced by the parasite species used at exposure on day 0 (χ 2 = 6.4, d.f. = 1, P = 0.012). Following both exposures, mean metacercarial load of individuals exposed to Echinostoma on day 0 was 30% higher than mean metacercarial load of individuals exposed to Echinoparyphium on day 0, regardless of day 3 treatment. When comparing expected infection loads with observed infection loads by treatment, mean metacercarial load of individuals exposed to Echinoparyphium on day 0 and Echinostoma on day 3 was 32% lower than the expected value (t = −3.98, d.f. = 9, P = 0.003; Fig. 3B). Observed infection loads did not differ significantly from the expected value in the other three treatments (all P > 0.05; Fig. 3B). We found no significant differences in mean metacercarial load between any of the treatments receiving only 25 total cercariae that were used to calculate expected infection values (χ 2 = 2.7, d.f. = 3, P = 0.436). Given that only four tadpoles died during the experiment, we did not assess the effects of treatment on tadpole mortality.
Kidney encystment bias
Of the 4375 cercariae used across all treatments, 1816 successfully encysted (41.5%). Of those that successfully encysted, 715 encysted in the left kidney (39.4%) and 1101 encysted in the right kidney (60.6%), and a repeated G-test of goodness-of-fit found that the bias toward the right kidney was statistically significant (G = 82.7, d.f. = 1, P < 0.001). The bias of metacercariae toward the right kidney was consistent within all seven treatments that received 50 cercariae (P < 0.05).
Discussion
In natural systems, hosts are typically challenged by a wide array of co-occurring parasites (Rynkiewicz et al., Reference Rynkiewicz, Pedersen and Fenton2015). Accordingly, disease ecologists have shifted focus from studies of single parasite infection outcomes to an assemblage of two or more parasites to better understand the influence of co-infection in disease outcomes, within-host parasite communities, and community disease dynamics (Telfer et al., Reference Telfer, Lambin, Birtles, Beldomenico, Burthe, Paterson and Begon2010; Hoverman et al., Reference Hoverman, Hoye and Johnson2013; Ezenwa, Reference Ezenwa2016; Wuerthner et al., Reference Wuerthner, Hua and Hoverman2017). While this prior research has demonstrated that coinfection with two functionally different parasites can have profound impacts on infection outcomes, there has been little focus on infection outcomes associated with coinfection by functionally and morphologically similar species (Miura et al., Reference Miura, Kuris, Torchin, Hechinger, Dunham and Chiba2005; Detwiler et al., Reference Detwiler, Bos and Minchella2010, Reference Detwiler, Zajac, Minchella and Belden2012). Using molecular methods, we found that two echinostome species, E. trivolvis and Echinoparyphium lineage 3, commonly co-occur in aquatic ecosystems. Using this qualitative field survey to inform our experimental design, we demonstrated that both single and sequential exposures to these two echinostome species can alter infection loads in a larval amphibian host.
In treatments that experienced a single exposure to two echinostome species – separately or combined – we found that mean metacercarial load of individuals exposed to a mixed cohort of Echinostoma and Echinoparyphium was 58% higher than the mean metacercarial load of individuals exposed to only Echinoparyphium but similar to the mean metacercarial load of individuals exposed to only Echinostoma. Individuals exposed to only Echinostoma had 49% higher infection loads than individuals exposed only to Echinoparyphium, but this difference was not statistically significant. Thus, co-occurring echinostome species can be highly variable in their ability to successfully infect larval amphibians and simultaneous infection with multiple species can cause infection loads equal to or higher than either species alone.
One potential mechanism for this observation is that in coinfection, one parasite facilitates increased infection success in the other. Because the two parasite species were present at the same time, it is unlikely that an immune response played a significant role in facilitation. Instead, infection by the more successful parasite may have increased the infection success of the other parasite by enhancing its ability to enter the host. For example, cellular damage caused by Echinostoma encystment may have increased the concentration of chemical cues released from the cloaca (e.g. amino acids; Haas et al., Reference Haas, Haberl, Korner, Spengler, Hertel and Kalbe2000), thereby increasing the ability of Echinoparyphium to enter the host. Alternatively, hosts exposed to only Echinoparyphium on day 0 may have had lower infection loads due to density-dependent regulation resulting from intraspecific interference or a stronger behavioural response by the host (Ebert et al., Reference Ebert, Zschokke-Rohringer and Carius2000; Karvonen et al., Reference Karvonen, Paukku, Valtonen and Hudson2003; Poulin, Reference Poulin2010). There may be an evolved mechanism for parasites to avoid penetration of previously infected hosts, although this has mainly been explored with the miracidia stage rather than the highly abundant cercaria stage (Haas et al., Reference Haas, Haberl, Korner, Spengler, Hertel and Kalbe2000; Allan et al., Reference Allan, Rollinson, Smith and Dunn2009; Vannatta et al., Reference Vannatta, Knowles, Minchella and Gleichsner2020). Because the metacercariae of the two species used are morphologically indistinct, it remains uncertain how coinfection influenced the success of each species alone; however, the patterns in this experiment suggest that infection success can differ between echinostome species but that simultaneous coinfection may increase the infection success of one or both species.
In treatments that experienced parasite exposures at two separate time points, we found that the mean metacercarial load of individuals exposed to Echinoparyphium on day 0 and Echinostoma on day 3 was 37% lower than either treatment with exposure to Echinostoma on day 0. We also found that individuals exposed to Echinoparyphium on day 0 had a 23% lower infection load relative to individuals exposed to Echinostoma on day 0. Together, these results demonstrate that priority effects can influence the overall infection success of two co-occurring echinostome species, but that the interaction may be asymmetric and dependent on exposure order. These results are unlikely to be explained by the energetic demands of the previously encysted parasites, as metacercariae have low resource demands (Smyth and Halton, Reference Smyth and Halton1983). Instead, the finding that initial exposure to Echinoparyphium resulted in overall lower infection loads than initial exposure to Echinostoma could indicate that either: (1) Echinoparyphium exposure induces a strong immune response that confers cross-immunity to a later echinostome infection; or (2) Echinostoma exposure results in helminth-induced immunosuppression that prevents cross-immunity to a later echinostome exposure from occurring. Helminth-induced immunosuppression is a well-documented strategy used by trematodes to increase their survival in the host (Maizels et al., Reference Maizels, Pearce, Artis, Yazdanbakhsh and Wynn2009; Taylor et al., Reference Taylor, van der Werf and Maizels2012), but it is unclear if this occurs in the metacercarial stage that infects amphibians. There is, however, evidence to suggest that early exposure to trematode cercariae can heighten resistance to a later challenge by a functionally different parasite (Hoverman et al., Reference Hoverman, Hoye and Johnson2013; Wuerthner et al., Reference Wuerthner, Hua and Hoverman2017; Koprivnikar et al., Reference Koprivnikar, Hoye, Urichuk and Johnson2019). The fact that mean infection load of individuals exposed to Echinoparyphium on day 0 and Echinostoma on day 3 was significantly lower than the expected value supports the cross-immunity hypothesis and suggests that the E. trivolvis on day 3 had an infection success lower than what would be expected if the infection was additive (i.e. antagonism). Collectively, these results suggest that exposure timing and order can strongly influence coinfection outcomes, even with functionally similar parasites; however, more research is needed to reveal the mechanism underlying these priority effects.
A limitation of this study is that in coinfected tadpoles, we were unable to differentiate between the cysts of two echinostome species. Several studies of amphibian trematode infections have used fluorescing dye to label cercariae used for exposures on different days, a potentially useful method for future studies of infection dynamics with morphologically similar parasites (Leung and Poulin, Reference Leung and Poulin2011; Hoverman et al., Reference Hoverman, Hoye and Johnson2013; LaFonte and Johnson, Reference LaFonte and Johnson2013; Johnson et al., Reference Johnson, Koprivnikar, Orlofske, Melbourne and Lafonte2014; LaFonte et al., Reference LaFonte, Raffel, Monk and Johnson2015; Koprivnikar et al., Reference Koprivnikar, Hoye, Urichuk and Johnson2019). Another limitation is that because our field surveys were limited in scope, we can draw few conclusions about how the prevalence of each species changes through the year in natural systems. For example, a temporal gap in peak abundance of Echinostoma and Echinoparyphium lineage 3 could lead to predictably staggered infections similar to those explored experimentally here. Alternatively, spatial separation of snails infected with different echinostomes within a pond could lead to temporal gaps in exposure based on host habitat choices. Further standardized surveys will be needed to determine if this is the case. Finally, because echinostome infection loads can significantly vary in natural situations, our narrow range of exposures likely does not capture the range of outcomes, such as increased pathology or mortality, that might occur with echinostome coinfections. Future studies should explore how coinfection with different echinostomes at a range of exposure loads alters disease outcomes and the success of each parasite.
As the principles of community ecology become increasingly important to the field of disease ecology, clarifying the dynamics of within-host parasite interactions has become necessary (Pedersen and Fenton, Reference Pedersen and Fenton2007; Johnson et al., Reference Johnson, De Roode and Fenton2015). Moreover, molecular studies continue to reveal genetic distinctions between morphologically similar species. Thus, it is important to explore whether there are non-additive interactions between functionally and morphologically similar macroparasite species or whether the effects of parasite exposure can be generalized by morphotype. Our study demonstrates that echinostome infection success in larval anurans can differ significantly based on the species makeup, density and timing of exposure. We also found evidence for priority effects based on exposure order; individuals exposed to Echinoparyphium on day 0 tended to have lower final infection loads than individuals exposed to Echinostoma on day 0. This finding adds to the existing literature demonstrating priority effects during coinfection and emphasizes that priority effects can occur even between functionally similar species (Hoverman et al., Reference Hoverman, Hoye and Johnson2013; Devevey et al., Reference Devevey, Dang, Graves, Murray and Brisson2015; Wuerthner et al., Reference Wuerthner, Hua and Hoverman2017). Given these findings, we recommend that workers using field-collected echinostomes as a model parasite for disease studies use molecular methods to confirm which species will be used. While the cost of sequencing can be prohibitive, techniques such as qPCR or PCR with species-specific primers (e.g. Fujino et al., Reference Fujino, Zhiliang, Nagano, Takahashi and Fried1997) may provide a more affordable way to rapidly identify the presence or absence of DNA from a specific species. Studies should continue to focus on how cryptic parasite diversity in natural systems influences disease outcomes, and how these functionally similar organisms interact and compete for within-host resources.
Acknowledgments
We thank J. Jaeger, N. Buss, B. Mattes, E. Yates, T. DeBlieux and D. DiGiacopo for their invaluable help in conducting the field surveys associated with this study. We also thank T. DeBlieux and E. Flaherty for their comments on earlier drafts of the manuscript.
Financial support
This research was supported by National Science Foundation grant #1655156 awarded to J.T.H., National Science Foundation grant #1655190 awarded to J.H. and National Science Foundation grant #1655168 awarded to R.A.R.
Conflict of interest
The authors declare no conflicts of interest.
Ethical standards
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional guides on the care and use of laboratory animals. The authors have involved the minimum number of animals to produce statistically reproducible results. All methods for the infection of tadpoles were approved by the Purdue University IACUC (protocol 1701001530). Animals were collected under Indiana Department of Natural Resources permit 18-066.
