Differential impacts of shared parasites on fitness components among competing hosts

Abstract Effects of parasites on individual hosts can eventually translate to impacts on host communities. In particular, parasitism can differentially affect host fitness among sympatric and interacting host species. We examined whether the impact of shared parasites varied among host species within the same community. Specifically, we looked at the impacts of the acanthocephalan Acanthocephalus galaxii, the trematodes Coitocaecum parvum and Maritrema poulini, and the nematode Hedruris spinigera, on three host species: the amphipods, Paracalliope fluviatilis and Paracorophium excavatum, and the isopod, Austridotea annectens. We assessed parasite infection levels in the three host species and tested for effects on host survival, behavior, probability of pairing, and fecundity. Maritrema poulini and C. parvum were most abundant in P. excavatum but had no effect on its survival, whereas they negatively affected the survival of P. fluviatilis, the other amphipod. Female amphipods carrying young had higher M. poulini and C. parvum abundance than those without, yet the number of young carried was not linked to parasite abundance. Behavior of the isopod A. annectens was affected by M. poulini infection; more heavily infected individuals were more active. Paracorophium excavatum moved longer distances when abundance of C. parvum was lower, yet no relationship existed with respect to infection by both M. poulini and C. parvum. The differential effects of parasites on amphipods and isopods may lead to community‐wide effects. Understanding the consequences of parasitic infection and differences among host species is key to gaining greater insight into the role of parasite mediation in ecosystem dynamics.

Furthermore, different parasite species can vary greatly in their impacts on hosts. For instance, the pairing success of male

Gammarus pulex amphipods was affected differently by infection with
Pomphorhynchus laevis than Polymorphus minutus, both acanthocephalan parasites (Bollache et al., 2001). Many parasite species are known to infect a variety of hosts, many of which may compete strongly with each other within their ecosystem. Differential impacts of parasites on competitors will affect their relative competitive abilities (Price et al., 1986). As species vary in their susceptibility and tolerance to parasites, the presence or absence of parasite species may dictate the coexistence of species or the complete absence of a species within an ecosystem (Greenman & Hudson, 2000;Hatcher, Dick, & Dunn, 2006). If two species are equal competitors, the presence of a parasite that infects only one of them may change this interaction. If the host is negatively affected by the parasite, it may give the competitor the advantage. The more tolerant host may also be able to act as a reservoir for the parasites, maintaining a high level of parasitism within the system (Arneberg, Skorping, Grenfell, & Read, 1998). To better understand the outcomes of competition and predation in the presence of shared parasites, it is important to understand the different impacts parasites may have on particular hosts.
Interactions among parasites may affect their respective virulence and the survival of the host (Alizon, 2013;Balmer, Stearns, Schötzau, & Brun, 2009;Lange et al., 2014;de Roode, Culleton, Cheesman, Carter, & Read, 2004). The overall virulence of a combination of parasites can be higher than that of the most virulent parasite, lower than the least virulent one or reach some intermediate level . Yet, despite the importance of interactions among shared parasites and multispecies infections, little is understood about how shared parasites may shape host communities.
The objectives of our study were to examine potential impacts of different parasite species on three species of hosts in the same community. We examined the effects of parasites on invertebrates used as intermediate hosts by four parasite species, all of which are transmitted trophically to their definitive host. Parasite effects on host fecundity, behavior, and survival have been previously examined in one of the host species, the amphipod Paracalliope fluviatilis (Lagrue & Poulin, 2008a;Rauque et al., 2011). It serves as host to two trematode species, Coitocaecum parvum and Maritrema poulini, and the acanthocephalan Acanthocephalus galaxii. Coitocaecum parvum, and A. galaxii use fish as definitive hosts while M. poulini is an avian parasite (Hine, 1977;MacFarlane, 1939;Presswell, Blasco-Costa, & Kostadinova, 2014).
Less is known about the impacts of these parasites on other hosts in the community. Additionally, potential effects of multiple infections are not well understood. Two other crustacean species are commonly found coexisting with P. fluviatilis and serve as hosts to some of the same parasites. Paracorophium excavatum, another amphipod, is larger than P. fluviatilis but they both occur in sympatry (Ruiz-Daniels, Beltran, Poulin, & Lagrue, 2012). Paracorophium excavatum is also host to three parasites, including the trematodes C. parvum and M. poulini, but also the fish nematode Hedruris spinigera (Lagrue & Poulin, 2008b;Luque, Bannock, Lagrue, & Poulin, 2007;Luque et al., 2010;Ruiz-Daniels et al., 2012). The prevalence and abundance of C. parvum and M. poulini have previously been found to be higher in P. excavatum than in P. fluviatilis (Ruiz-Daniels et al., 2012). The isopod Austridotea annectens is also found in the same area and is an intermediate host for M. poulini Presswell et al., 2014). Many of these parasites reach a relatively large size and/or abundance within their hosts, suggesting potential impacts on host survival and behavior (Rauque et al., 2011). Our specific objectives were to (i) determine whether parasite effects varied among hosts within the same community and (ii) test whether multispecies infections had synergistic or antagonistic effects compared to the presence of single parasite infection Lagrue & Poulin, 2008a;Lange et al., 2014).
As these hosts are all competing for resources and share a variety of parasites, a better understanding of the impacts of parasitism on each host is necessary to understand how parasites affect population dynamics in this community.

| Sample collection
We collected samples of naturally infected amphipods and isopods from the littoral zone of Lake Waihola, South Island, New Zealand (46°01′14S, 170°05′05E) between February and September 2016.
Sampling for survival tests occurred over 3 days, 9 February, 21 March, and 1 May 2016. Sampling for the behavioral tests occurred over 3 days, 13 May, 1 June, and 6 September 2016, due to seasonal variation in host abundance. Sampling for pairing behavior occurred on 21 March 2016 for P. fluviatilis (as described below) but paired isopods were collected during the entire sampling period as they occurred far less frequently. Data on the size, sex, and the prevalence and abundance of parasites in each host species from each sample event were pooled for examination of intraspecific and interspecific variation. Animals were caught using dip-nets and transported to the laboratory in lake water. Amphipods and isopods were transferred and maintained separately by species in 10 L tanks containing aerated lake water. Animals were kept at room temperature (14 ± 1°C) with aquatic plants (Myriophyllum triphyllum and Elodea canadensis) for food, and under a controlled photoperiod (12-hr dark and light).
All amphipods used in behavioral and fecundity trials were subsequently dissected within a week of collection as keeping these amphipods in the laboratory for long periods of time can affect amphipod survival Lagrue, Poulin, & Keeney, 2009;Poulin, 2003). All isopods used in behavioral trials were dissected within a month of collection. If individuals did not die during trial, they were killed in 70% ethanol and rinsed in distilled water before dissection. In our study, prevalence was defined as the percentage of infected hosts, abundance was defined as the number of parasites per host including zeroes, and mean abundance as the mean number of parasites among a specific sample of hosts.

| Survival tests
Within 6 hr after sampling, 266 P. fluviatilis, 210 P. excavatum, and 390 A. annectens were separated into individual wells of tissue culture plates. According to host size, P. fluviatilis were maintained in 96-well microplates with 300 μL of water per well, P. excavatum were maintained in 24-well plates with 500 μL of water per well, and A. annectens were kept in 12-well plates with 1 mL of water per well. All individuals were maintained at the same temperature (14 ± 1°C) and photoperiod (12-hr dark and light) but no food was added. Well plates were checked daily for any dead individuals. If a female released young, the number of young was recorded and they were removed from the well as they could have provided an additional food resource to the focal animal through cannibalism. If individuals could be dissected the same day, they were left in lake water. For dissection occurring more than 24 hr after death, the individuals were immediately preserved in 70% ethanol until dissection. The total body length of each individual was determined by measuring from the anterior tip of the cephalic capsule to the posterior end of the uropods. Sex was determined for each individual when possible. Isopods shorter than 7-8 mm in body length were impossible to sex due to the lack of secondary sexual characters and thus considered juveniles. Egg presence and number were also recorded. Individuals were then dissected to identify and count parasites.

| Behavioral tests
Amphipods and isopods were individually isolated (in wells of culture plates, as described above) within 6 hr of sampling and left for 12 hr to acclimate to their new environments. Individuals were subsequently filmed to record velocity and activity levels. Infection status of each individual was unknown during filming and subsequent video analysis.
Fine sand was added to the wells of P. excavatum and A. annectens to simulate natural conditions as both species are benthic and may use sand to burrow. One hour prior to recording, plates were moved to the filming studio to allow the animals to adjust to the temperature (18 ± 1°C) and lighting changes. Paracalliope fluviatilis and P. excavatum were filmed for 5 min under a dissecting microscope (Olympus SZ61, 0.65× magnification) due to their small size. Well plates containing A. annectens were filmed for 5 min using a Canon digital camera (1200D). Activity levels (distance moved from center (mm), mean velocity (mm/ms), highly mobile duration (more than 60% of the animal (measured by pixels altered) has moved within the sample period (0.05 s), which is then calculated as a proportion of the entire sample to give a measure of high-speed movement), and mobile duration (more than 20% of the animal (measured by pixels altered) has moved within the sample period (0.05 s), which is then calculated as a proportion of the entire sample to give a measure of movement)) were calculated for each individual over 5 min using EthoVision XT

| Fecundity and pairing probability
Offspring carried in the brood pouch of gravid females (from the survival and behavioral tests) were counted and recorded with their corresponding body length and parasite burdens. If any young were released during survival tests, the number of young was recorded and matched with the female's corresponding parasite burden upon death.
The parasite burden of females without any young was also compared to those with young to examine whether a relationship existed between the probability of having young and a female's parasite burden.
Additionally, a subset of paired P. fluviatilis and A. annectens were identified and individually separated into tissue culture microtest tubes within 12 hr of capture. The paired individuals consisted of a male clasping a female in a precopulatory pair. Nonpaired individuals were also collected during the same sampling event and separated into individual tubes. Amphipods were dissected as described above within 24 hr of capture. Isopods were euthanized and preserved in ethanol and dissected as described above.

| Statistical analysis
Statistical analyses were performed in JMP ® 12 (SAS Institute Inc 2015) and R statistical software (http://www.R-project.org). Size differences between sexes were examined using a Kruskal-Wallis test.
The relationship between host size and parasite abundance was assessed using Spearman's correlations. Differences in parasite abundance and prevalence among host species and between sexes of each individual parasite species were examined using a Kruskal-Wallis test and contingency analysis, respectively. The relationship between either survival (days before death) or behavioral measures, and parasite abundance of C. parvum, M. poulini, A. galaxii, and H. spinigera, as well as multispecies combinations, was analyzed separately for each host species using a negative binomial regression, with host size being included as an additional explanatory variable and their interactions when significant. Logistic regression was used to relate parasite abundance with both the likelihood of being paired in P. fluviatilis and A. annectens by sex and the probability of having young in female P. fluviatilis and P. excavatum. The number of young carried by a female was related to its body size and parasite abundance using negative binomial regression.
A synopsis of the key results of this study is summarized in Table 2.

| Host survival and parasite infection
The survival (i.e., number of days before death) of P. fluviatilis was negatively related to M. poulini and C. parvum abundance and host size (Table 4, Figure 1). There was no interactive effect between either parasite abundance and size (  Figure 1).
Intriguingly, there was a positive relationship between M. poulini abundance and survival in A. annectens, as well as a relationship between host size and parasite abundance with an interactive effect between size and abundance (Table 4, Figure 1).

| Host behavior
The distance moved from the center in P. fluviatilis depended on the abundance of A. galaxii (Table 4). While all other measures of behavior in P. fluviatilis (velocity, high mobile duration, and mobile duration) were unaffected by parasite abundance (A. galaxii, M. poulini, or C. parvum).
Paracalliope fluviatilis size did influence host behavior, as larger hosts moved further and had longer highly mobile durations (Table 4).
In P. excavatum, the distance moved from the center of the well plate was negatively related to the abundance of C. parvum with a significant interactive effect between host size and C. parvum abundance  (Table 4). Additionally, individuals who spent more time highly mobile tended to have a lower abundance of C. parvum or to be smaller, with a trend toward an interactive effect between C. parvum and size (Table   4). Mobile duration (time spent moving) of P. excavatum was negatively related to size and tended to be positively related to M. poulini abundance with a trend toward an interactive effect between M. poulini abundance and size (Table 4). However, there was no relationship between or distance moved or high mobile duration and the abundance of M. poulini, H. spinigera, or size (Table 4). Mobile duration was not related to the abundance of A. galaxii or that of C. parvum. Velocity (mm/ ms) was not related to the abundance of any parasite or size (Table 4).
Mobile duration (time spent in motion) of A. annectens was negatively related to M. poulini abundance but not to host size, although there was a significant interaction effect (Table 4, Figure 3). High mobile duration was not related to M. poulini abundance, but it was related to size (Table 4). No relationship was found between velocity (ms/s) or distance moved from center and M. poulini abundance or size (Table 4). A sample of paired and unpaired P. fluviatilis (Table 3) was used to examine potential parasite effects on the likelihood of pairing in both sexes. The abundance of A. galaxii was higher in paired females than unpaired females (Table 3). Additionally, the prevalence of C. parvum was higher in nonpaired females compared to paired females (Table 3).

| Fecundity and pairing probability
No other differences were found in parasite abundance between paired and unpaired female and male P. fluviatilis (including multiple infections, Table 3).
Paired and unpaired A. annectens were used to examine the possible influence of parasites on the likelihood of pairing. No difference in M. poulini abundance was found in either males or females between paired and unpaired individuals (Table 3). However, there was a trend for a higher prevalence of M. poulini in paired male isopods compared to nonpaired individuals (Table 3).

| DISCUSSION
Understanding impacts of shared parasites on a variety of sympatric host species is necessary for making predictions on the potential Infection levels varied among hosts within the community.
Paracorophium excavatum had higher abundance of M. poulini than P. fluviatilis (Table 1), which is consistent with previous studies (Presswell et al., 2014;Ruiz-Daniels et al., 2012). Variation in host size or other biological characteristics may lead to these differences (Grutter & Poulin, 1998;Johnson, Bush, & Clayton, 2005;Ruiz-Daniels et al., 2012;Saad-Fares & Combes, 1992). Paracorophium excavatum is a larger amphipod that has a more benthic habitat than P. fluviatilis as it is often found burrowing in the sand. However, we only found a positive correlation between the size of P. excavatum and one species of parasite, contrary to previous studies (Presswell et al., 2014;Ruiz-Daniels et al., 2012). The lack of difference in prevalence and abundance of C. parvum between amphipod species may have been due to the low overall abundance and prevalence of C. parvum within the community, making it more difficult to detect any effect of this parasite on its hosts. Additionally, as discussed more thoroughly below, we found no relationship between the abundance of M. poulini and the survival of P. excavatum, suggesting that interspecific differences in abundance between the two amphipods may be due to higher parasite-induced mortality in P. fluviatilis following infection and/or parasite accumulation.
The impact of parasites on survival varied among the three host species. The amphipod P. fluviatilis was shown to have reduced survival when infected with both C. parvum and M. poulini (Table 4, Figure 1). Individuals with multispecies infections also had a lower survival than individuals with no parasites. Reduced survival incurred by hosts may be due to energetic costs of infection. It has been previously suggested that the negative effect of M. poulini (previously referred to as Microphallus sp.) on P. fluviatilis is a direct consequence of high parasite abundance (Rauque et al., 2011). Our results suggest that parasite-induced mortality may influence differences in parasite abundance between amphipod species rather than size differences between the two host species. Paracorophium excavatum survival was not impacted by the abundance of any of its three parasites (Table 4, Figure 1). The stark contrast in response to similar parasites by the two amphipod species may be due to a difference in the virulence of the parasite between hosts as documented in several prior studies (Jensen, Thomas Jensen, & Mouritsen, 1998;Park, 1948;Thomas et al., 1995). The positive relationship between M. poulini abundance and survival in A. annectens was surprising although it may be an effect of host size (Table 4, Figure 1). The abundance of M. poulini in this isopod may be directly linked to exposure over time and as age is linked to size, a relationship between both is not surprising. Additionally, larger individuals may be able to survive longer, due to a higher resource base to draw upon, therefore creating the appearance of a positive relationship between parasite abundance and survival.
Hosts exhibited different behavioral responses related to the abundance of parasites. Interestingly, P. fluviatilis only demonstrated behavioral changes when infected with A. galaxii. We did not observe behavioral effects of the other parasite species as seen in a previous study (Rauque et al., 2011), and observations of parasite-induced behavioral modification in this amphipod species have been mixed (Poulin, 2001). The inconsistency with the first study could be due to the behavioral aspects measured (phototaxis versus activity levels) (Rauque T A B L E 3 Parasite abundance (mean ± SE) and prevalence (percentage of sample infected) in paired and unpaired Paracalliope fluviatilis and Austridotea annectens (listed with sample size per group). All samples were collected from Lake Waihola, South Island, New Zealand between February and September 2016  (Rauque et al., 2011). The negative relationship between movement and abundance of C. parvum appears nonadaptive for the parasite, as increased movement can increase risk of predation, allowing C. parvum to be passed on to its fish definitive host, mainly the common bully (Gobiomorphus cotidianus). The common bully can use nonvisual methods, such as olfactory, tactile, or lateral-line prey detection, to find prey and the lack of moment in Paracorophium excavatum may allow them to avoid being depredated by this fish (Rowe, 1999;Rowe, Nichols, & Kelly, 2001). The decrease in activity levels may be due to reduced energy available due to parasite infection.
Parasites had intriguing impacts on the fecundity and probability of pairing in the various hosts. Female P. fluviatilis and P. excavatum carrying young had a higher abundance of M. poulini. Previous studies have indicated that "handicapped' P. fluviatilis females, that is, with artificially impaired swimming performance, were more likely to be paired (Sutherland, Hogg, & Waas, 2007), and therefore, females with parasites may also have been easier to pair with, explaining the higher likelihood of having young in females with a higher parasite abundance.
However, the abundance of both C. parvum and M. poulini was not linked with the probability of male or female P. fluviatilis being found in pairs, contrary to previous studies (Rauque et al., 2011). An alternative hypothesis may suggest that females may choose to invest more into reproduction while infected; therefore, more infected female amphipods would be carrying young (Agnew, Koella, & Michalakis, 2000).
Yet, we did not find any relationship between the number of young carried and the abundance of any parasites.
Multispecies infections have important implications for the fitness of individuals and the dynamics of populations. In our study T A B L E 4 Negative binomial regression models for relative host survival and behavior measures compared to parasite infection and host size. All samples were collected from Lake Waihola, South Island, New Zealand between February and September 2016  (Cezilly, Gregoire, & Bertin, 2000). As multispecies infections are more common in P. excavatum than P. fluviatilis and appear to potentially be more important to the impacts of parasites on this host species, these co-infection effects may have impacts on the host population overall.
Our study was based on natural infections rather than an experimental approach thus limiting our interpretations, as we cannot directly address causality or the direct mechanisms of these interactions.
However, if combined with what is known of the mechanistic bases of parasite-induced host modifications (see Introduction section), it remains a strong approach for examining the effects of parasites on their hosts (Poulin, 2001).
Interestingly, the effects of parasites on survival and behavior varied greatly between amphipod hosts. Not only does P. excavatum have a much higher abundance and prevalence of M. poulini but there appears to be no decreased survival with a high abundance of either trematode parasite (Table 2). If P. fluviatilis and P. excavatum are competing for the same resources, this may give P. excavatum an advantage. A more tolerant host, such as P. excavatum, may also be able to act as a reservoir for the parasites, maintaining a high level of parasitism within the system (Arneberg et al., 1998). This may increase infection risk for P. fluviatilis with possible consequences for behavior, reproduction, and/or survival. The more tolerant species may become the stronger competitor within the ecosystem, and thus, the parasite could mediate the interaction between the two hosts and alter the outcome of competition (Greenman & Hudson, 2000).
Variation in how parasites affect their different host species has the potential to have community-wide effects. As our species vary in their susceptibility and tolerance to parasites, the presence or absence of a parasite species within a system may dictate the coexistence of species or the extirpation of a particular species (Begon, Bowers, Kadianakis, & Hodgkinson, 1992;Greenman & Hudson, 2000;Hatcher et al., 2006;Hudson, Dobson, & Lafferty, 2006). The differential effects on amphipods and isopods may lead to community-wide effects.
Understanding the consequences of parasitic infection and differences between host species is key to gaining greater insight into the role of parasite mediation in ecosystem dynamics.

ACKNOWLEDGMENTS
Financial support for this project was provided by the Department of Zoology, University of Otago. Special thanks to S. Johnson for providing access to EthoVision and assistance in its use. Field and laboratory assistance was provided by A. Filion, K. Karl, B. Presswell, R. Rinnevalli, Z. Tobias, and M. Villa. Special thanks to A. Stumbo, R.
Gigliotti, S. Jarvie, Z. Tobias, B. Presswell, and C. MacLeod for comments and advice on earlier drafts of this manuscript.

CONFLICT OF INTEREST
None declared.
F I G U R E 3 Host behavior measures compared to parasite abundance; (a) distance moved from center versus the abundance of Coitocaecum parvum in Paracorophium excavatum (n = 84; negative binomial regression, z = −2.0, p = .049) (b) mobile duration (time moving) of individual compared to their Maritrema poulini abundance in Austridotea annectens (n = 63; z = 2.88, p = .004). All measures were taken during a 5-min behavioral observation. Regression lines represent the direction of relationships. All animals were collected from Lake Waihola, New Zealand between February and September 2016.