Recolonizing gray wolves increase parasite infection risk in their prey

Abstract The recent recolonization of Central Europe by the European gray wolf (Canis lupus) provides an opportunity to study the dynamics of parasite transmission for cases when a definitive host returns after a phase of local extinction. We investigated whether a newly established wolf population increased the prevalence of those parasites in ungulate intermediate hosts representing wolf prey, whether some parasite species are particularly well adapted to wolves, and the potential basis for such adaptations. We recorded Sarcocystis species richness in wolves and Sarcocystis prevalence in ungulates harvested in study sites with and without permanent wolf presence in Germany using microscopy and DNA metabarcoding. Sarcocystis prevalence in red deer (Cervus elaphus) was significantly higher in wolf areas (79.7%) than in control areas (26.3%) but not in roe deer (Capreolus capreolus) (97.2% vs. 90.4%) or wild boar (Sus scrofa) (82.8% vs. 64.9%). Of 11 Sarcocystis species, Sarcocystis taeniata and Sarcocystis grueneri occurred more often in wolves than expected from the Sarcocystis infection patterns of ungulate prey. Both Sarcocystis species showed a higher increase in prevalence in ungulates in wolf areas than other Sarcocystis species, suggesting that they are particularly well adapted to wolves, and are examples of “wolf specialists”. Sarcocystis species richness in wolves was significantly higher in pups than in adults. “Wolf specialists” persisted during wolf maturation. The results of this study demonstrate that (1) predator–prey interactions influence parasite prevalence, if both predator and prey are part of the parasite life cycle, (2) mesopredators do not necessarily replace the apex predator in parasite transmission dynamics for particular parasites of which the apex predator is the definitive host, even if meso‐ and apex predators were from the same taxonomic family (here: Canidae, e.g., red foxes Vulpes vulpes), and (3) age‐dependent immune maturation contributes to the control of protozoan infection in wolves.


| INTRODUCTION
Apex predators play a critical role in shaping food webs (Estes et al., 2011). When a predator is a definitive host of a parasite and disappears from its habitat, it may leave a gap in the food web. Potential consequences of such disappearances for parasite-host relationships are still poorly understood. Even for well-studied temperate ecosystems, where the gray wolf (Canis lupus) is an apex predator and ungulates are its main prey, little is known about the parasitological consequences of a transient wolf removal/extinction (East, Bassano, & Ytrehus, 2011). Transmission dynamics of trophically transmitted pathogens and parasites that are well adapted to a specific host might change. Parasites could either adapt to alternative hosts or disappear overtime (Farrell, Stephens, Berrang-Ford, Gittleman, & Davies, 2015), which we call the "host flexibility" hypothesis and the "fading out" hypothesis, respectively. Wolves are definitive hosts for a wide range of endoparasites (Craig & Craig, 2005), but little is known about their possible influence on parasite prevalence in their ungulate prey if these serve as intermediate hosts, as in the case of helminths or apicomplexa . In particular, it is unclear whether infection risk increases when a definitive host returns after being absent from a specific area for some time, and how infection risk varies among different prey species. It is also unclear which factors control parasite etiopathology and whether these factors favor specialization of parasites for specific hosts. The process of the current wolf recolonization of Central Europe provides an excellent opportunity to investigate parasite transmission dynamics in a predator-prey system as the same prey species can be examined in the presence and absence of the predator in the same habitat type.
Host-pathogen interactions and epidemiology are well understood in apicomplexan taxa such as Toxoplasma-a parasite occurring within a domestic and a sylvatic cycle with zoonotic potential (Shaapan, 2016). However, the links between Sarcocystis of wild intermediate and definitive hosts are currently unclear, and the prevalence and distribution of Sarcocystis species in ungulates and the potential impact of the removal and then the return of the apex predator are at present unknown.
Free-ranging wolves from the CEL population host at least 12 different Sarcocystis species . The genus Sarcocystis has an obligatory two-host life cycle involving (partially) carnivorous definitive hosts and a broad range of intermediate hosts such as reptiles, birds, or mammals (Dubey & Lindsay, 2006;Munday, Hartley, Harrigan, Presidente, & Obendorf, 1979). Sarcocystis are known to be more host-specific in terms of their intermediate than definitive host range, although the current state of knowledge is far from complete, as new Sarcocystis species and new hosts continue to be described (Dahlgren & Gjerde, 2010;Gjerde, 2014b when Sarcocystis penetrate the digestive mucosa and migrate through the blood vessels to reach their target (muscular or nervous) tissue where they eventually form ((sarco-)cysts) (Dubey, Calero_Bernal, Rosenthal, Speer, & Fayer, 2015;Dubey & Lindsay, 2006;Poulsen & Stensvold, 2014). During the early infection phase, pathogenic species may cause clinical symptoms such as weight loss, anemia, fever, and abortion in pregnant intermediate hosts (Buxton, 1998;Dubey & Lindsay, 2006)-otherwise, sarcocystosis usually has an asymptomatic etiopathology and minor impact on its host.
In this study, we investigated the prevalence of Sarcocystis in ungulates and wolves. Under the "fading out" hypothesis, returning wolves would reimport temporarily faded parasites, thereby increasing parasite infection risk in ungulate intermediate hosts. It assumes that at least some Sarcocystis species are "wolf specialists" and are too host-specific to use alternative hosts as definitive hosts. If an increase in Sarcocystis prevalence occurred, it should therefore be driven by "wolf-specialized" parasites, that is, Sarcocystis species that are particularly well adapted to wolves. "Wolf-specialized" parasites should then be overrepresented in wolves and show the strongest prevalence increase in ungulate intermediate hosts in wolf-inhabited areas, and there should be a "mismatch" in relative parasite frequencies between wolves and their prey. Under the "host flexibility" hypothesis, returning wolves serve as an additional definitive host for endemic parasites also spread by other carnivores (spillback) which had resumed the function of alternative hosts (Kelly, Paterson, Townsend, Poulin, & Tompkins, 2009;Moré, Maksimov, Conraths, & Schares, 2016). In this case, we would not expect to find Sarcocystis species that could be considered "wolf specialists". Without "wolf-specialized" parasites, relative parasite frequencies in ungulate prey species and in wolves would match as wolves would be nonselectively infected with the Sarcocystis they consume.
The "fading out" hypothesis also predicts that if parasites are particularly well adapted to a specific host ("wolf specialists"), we would expect them to prevent clearance by the host immune system. Young wolves are likely to be more susceptible to apicomplexan Sarcocystis parasites than older animals (in terms of less previous parasite exposure, gradual building up of immunity and hence a generally weaker immune response). Younger wolves would therefore be expected to exhibit a higher Sarcocystis species richness than adult wolves. In adults, an improved immune competence should allow them to clear parasites that might infect pups, except for "wolf specialists" which might employ adaptations that allow them to circumvent the host immune system and persist in older individuals. If no age-related immune processes controlled parasite resistance in wolves, wolves of all ages should host the same Sarcocystis community as each pack member is exposed to the same Sarcocystis species when they share an infected kill. were examined for the presence of intestinal Sarcocystis spp. Wolves were collected as roadkills or as confiscated poached animals originating from five federal states in northern and eastern Germany (50°10′-54°54′ N and 6°41′-15°2′ E). Wolf age classes (pup, yearling, adult) were determined as previously described .

| DNA extraction, PCR, and library preparation
Wolf intestinal contents were extracted and processed using the amplicon sequencing approach described in Lesniak, Heckmann et al.,

| Bioinformatics
Ungulate Sarcocystis sequences were sorted into operational taxonomic units (OTUs) using USEARCH (Edgar, 2010(Edgar, , 2013, which then were assigned to Sarcocystis species as described in Lesniak, Franz et al. (2017). Briefly, OTUs were assigned to Sarcocystis species sequences from a custom database (Lesniak, Franz et al., 2017) using BLAST ® (blastn; Altschul et al., 1990) with an identity threshold of 98%. Only hits with a biunique best bit score for one species were collected in a table including the respective Sarcocystis species, OTU, amplicon, and sample name. Due to technical limitations, the proti15_proti440_R1 and proti15_proti440_R2 datasets were split by ungulate species. When describing and discussing our results, the term "species" instead of "OTUs" will be used for simplicity, although we are aware of the technical limitations of our approach to determine species, as previously discussed .
The wolf metabarcoding dataset was analyzed as previously described .

| Statistical analyses
The data on Sarcocystis presence in ungulate tissues collected using light microscopy were used to test the prediction that ungulate Sarcocystis prevalence was higher in areas affected by wolf recolonization. Generalized linear models (GLMs) were fitted separately for each ungulate species with binomially distributed errors, in which the response variable was the record of "Sarcocystis spp. infection" (binary: infected or not infected). All models included the predictors "wolf presence" (binary: absent or present) and ungulate age (categorical: juveniles, subadults, adults). To scale the Sarcocystis infection risk of ungulates depending on the study site, we calculated the odds ratios for each ungulate species as an exponential function of the coefficient "wolf presence" extracted from the respective GLM.
In order to interpret the goodness of fit of each model in comparison with the null model, overall likelihood ratio tests were performed with the R package lmtest v0.9-34 (Zeileis & Hothorn, 2002), and model predictors were tested for collinearity using the R package car v2.6-26 (Fox & Weisberg, 2011).
In order to test whether some parasites occurred more frequently in wolves than expected, considering the frequencies of these parasites in the prey species, we first estimated expected frequencies in wolves (f exp ), taking into account that prey species are not necessarily consumed in equal proportions. The expected frequencies were then compared to observed parasite frequencies (f obs ) from wolves collected in this study ( Figure 2a) to (1) test whether a "mismatch" could be detected in terms of a significant difference in both distributions, and (2) identify which Sarcocystis spp., if any, were overrepresented in wolves.
Two approaches (A and B) were used to estimate expected parasite frequencies f exp in wolves. In approach A, the conventional approach, f exp was estimated based on the published information on wolf diet (Wagner et al., 2012) to derive the proportion of each prey species in the diet (feeding proportion d j ) and information on relative parasite infection frequencies p i,j in ungulates obtained in this study.
The observed relative frequencies pi,j of each Sarcocystis species i in ungulate species j in our sample are listed in Table 1. Based on published information on wolf diet, we used the following feeding proportions d j : red deer: 0.22, roe deer: 0.59, and wild boar: 0.19. Expected frequencies of Sarcocystis species i in wolves were then calculated as: The conventional approach has two major drawbacks. Firstly, identified mismatches between observed and expected parasite infections in wolves could be an artifact of erroneously estimated feeding proportions d j. Published information on the average wolf diet is not necessarily an accurate representation of individuals in this study, and therefore d j estimated in this way might be a poor representation of the diet of the wolves we analyzed. Secondly, it accounts for neither the number of preyed individuals nor the individual degree of prey parasite infestation.
To account for this potential problem, we also used a conservative approach (B) that aimed to minimize the chance of obtaining an erroneously elevated mismatch between observed and expected parasite frequencies in wolves. For this purpose, we indirectly es- Hochberg procedure with a false discovery rate of 5% (Benjamini & Hochberg, 1995). Overrepresented Sarcocystis species were considered candidate species for "wolf specialists," whereas all other species were termed "non-wolf specialists." Using a Mann- Whitney U test, we tested for each ungulate prey species whether "wolf-specialized" Sarcocystis had a higher increase in prevalence (δ prevalence ) in wolf areas relative to control areas in comparison with other detected Sarcocystis spp.
The wolf metabarcoding dataset was used to test the prediction that young wolves had a higher Sarcocystis species richness than adults, and that "wolf-specialized" Sarcocystis species persist in the adult age class, whereas "nonwolf specialists" fade out. were performed using the R package multcomp v1.4-5 (Hothorn, Bretz, & Westfall, 2008). In a next step, we tested whether potential "wolf-specialized" Sarcocystis are more likely to persist with increasing wolf age than "non-wolf specialists" that should be cleared by (

| Ungulate Sarcocystis spp. infection status and prevalence
Microscopic examination revealed that Sarcocystis sp. prevalence in ungulates was consistently higher in wolf areas than in control areas ( Figure 1). S. grueneri and S. taeniata occurred significantly more often in wolves than expected (representing potential "wolf specialists"), whereas S. bovini, S. miescheriana, S. silva, and S. truncata occurred less often than expected (Figure 2a).
Under the more conservative approach B, the observed Sarcocystis spp. infection frequencies in wolves also differed from expected probabilities (χ 2 = 78.067, df = 10, p < .0001). Post hoc tests showed that the same species, S. grueneri and S. taeniata, were overrepresented in wolves, suggesting they were "wolf specialists," whereas S. bovini, S. silva, and S. truncata occurred significantly less often than expected ( Figure 2b).
We also tested in ungulates whether the increase in prevalence from control areas to wolf-inhabited areas of "wolf specialist" Sarcocystis was higher than the increase in the other Sarcocystis species. Consistent with the idea of S. grueneri and S. taeniata being "wolf specialists" identified by approach A and B, their increase in prevalence from control area to wolf-inhabited area was significantly higher than the increase in the other Sarcocystis species in both red deer and roe deer (red deer: U = 18, p = .044; roe deer: U = 18, p = .043).

| Sarcocystis spp. species richness in wolves and prevalence of "wolf specialists"
Eleven known Sarcocystis species were detected from wolf intestinal samples of which six were also detected in wild ungulates.
F I G U R E 1 Observed Sarcocystis sp. prevalence in three ungulate prey species in relation to the wolf presence in their habitat. Sarcocystis sp. prevalence was significantly higher in wolf areas (dark gray) than in control areas (light gray) in red deer (n WT = 75, n CA = 18, p < .001), corresponding to a 47.0 times increased Sarcocystis sp. infection risk in red deer from the wolf area. A nonsignificant trend was found for roe deer (n WT = 99, n CA = 72, p = .075) and wild boar (n WT = 83, n CA = 37, p = .097), corresponding to a 3.8 times and 2.2 times increased infection risk of roe deer and wild boar The mean value for the prevalence change of the "wolf specialists" S. grueneri and S. taeniata between wolf pups and adults was 19.5%. This expected value differed significantly from the prevalence difference of the three common nonspecialist species S. capreolicanis, S. gracilis, and S. miescheriana (one-sample t test, t = 5.885, C.L. 6.587-42.413, p = .028, Figure 4).

| DISCUSSION
In this study, we investigated whether the return of an apex predator affected parasite transmission dynamics in its ungulate prey, as measured by the prevalence of the apicomplexan genus Sarcocystis.
For most species, sarcocysts reside in muscles in the prey and is only F I G U R E 2 (a) Relative observed (dark gray) versus relative expected (light gray) Sarcocystis spp. infection frequencies in wolves based on approach A. Expected values were generated by counting genetically determined Sarcocystis spp. occurrences in red deer, roe deer, and wild boar and by accounting for relative, normalized ungulate feeding proportions by wolves extracted from the literature (Wagner et al., 2012). The general distribution of infection probabilities was significantly different between observed and expected values (chi-squared test for given probabilities, χ 2 = 120.47, df = 10, p < .0001). Binomial post hoc tests showed that Sarcocystis grueneri (p = .018) and Sarcocystis taeniata (p = .018) were overrepresented in wolves, whereas Sarcocystis bovini (p = .036), Sarcocystis miescheriana (p = .018), Sarcocystis silva (p = .018), and Sarcocystis truncata (p = .018) were underrepresented. (b) Relative observed (dark gray) versus relative expected (light gray) Sarcocystis spp. infection frequencies in wolves based on approach B. Expected values have been generated by counting genetically determined Sarcocystis spp. occurrences in red deer, roe deer, and wild boar and by accounting for estimated relative ungulate feeding frequencies by wolves. The general distribution of infection probabilities is significantly different between observed and expected values (chi-squared test for given probabilities, χ 2 = 78.067, df = 10, p < .0001). Binomial post hoc tests showed that S. grueneri (p = .036) and S. taeniata (p = .018) were overrepresented in wolves, whereas S. bovini (p = .036), S. silva (p = .018) and S. truncata (p = .018) were underrepresented transferred to the definitive host if the intermediate host is eaten by a susceptible definitive host. Apicomplexans such as Sarcocystis with a two-host life cycle are rarely studied in their definitive hosts due to methodological challenges Moré et al., 2016;Xiang et al., 2009). In intermediate hosts, morphological cyst characteristics have frequently been used to microscopically identify Sarcocystis species (Malakauskas & Grikienienė, 2002;Odening, Stolte, Walter, & Bockhardt, 1995). However, oocysts or sporocysts isolated from definitive host intestinal samples do not permit morphological discrimination of species (Khan & Evans, 2006;Stronen, Sallows, Forbes, Wagner, & Paquet, 2011). To overcome such limitations, we used a combination of classical microscopy and metabarcoding to investigate Sarcocystis fauna, distribution, and transmission dynamics between wolves and their ungulate prey. We documented that prevalence was higher in ungulate prey in wolf-inhabited areas than in control areas, identified S. grueneri and S. taeniata as "wolf specialists", and showed that Sarcocystis species richness in wolves declined with age, whereas well-adapted species persisted in adults.

| Sarcocystis infection in ungulates
The presence of wolves in our study site was associated with a general increase in Sarcocystis sp. prevalence in their prey. Specifically, red deer had a much higher Sarcocystis sp. prevalence when sharing their habitat with wolves than animals from the control area. A similar, albeit statistically insignificant, trend was observed in roe deer and wild boar. A study from the Baltic states, where wolves have been continuously present (Chapron et al., 2014), reported similarly high prevalences of between 84.2% and 89.1% in three ungulate species, including red deer (Malakauskas & Grikienienė, 2002). The wolfassociated prevalence differences measured in red deer in this study and the comparison to wolf range states show that wolves should be considered a more frequent and important apex predator and consumer of red deer and its sarcocysts than smaller carnivores such as red foxes (Vulpes vulpes) or other mesopredators. As a result, red deer-associated Sarcocystis spp. cycles have increased in amplitude (prevalence and individual parasite burden) after wolf recolonization, consistent with the "fading out" hypothesis. Nevertheless, there are Sarcocystis spp. which can also be spread by mesopredators as definitive hosts , as indicated by a high sarcocysts prevalence in wild boar and roe deer in the absence of wolves. During the late 1970s, a study of roe deer in Germany described a Sarcocystis prevalence of 71.8%, even though no wolves were present in Central F I G U R E 3 Sarcocystis species richness decreases with wolf age (n wolves = 43). Pups had a significantly higher Sarcocystis species richness than adults (p = .021) F I G U R E 4 Sarcocystis spp. prevalence in wolf pups (n = 20, dark gray) and adults (n = 12, light gray). Eleven Sarcocystis species were detected in wolf intestinal samples, of which six species are shared with their wild ungulate prey species, including the "wolf specialists" Sarcocystis taeniata and Sarcocystis grueneri, whereas the other parasite species are known to have a (semi)domestic life cycle and did not occur in the investigated wild ungulates in this study. "Wolf specialists" had a lesser decrease in prevalence from pups to adults relative to "nonspecialist" Sarcocystis spp. (one-sample t test, t = 5.885, C.L. 6.587-42.413, p = .028) Europe at that time, which is consistent with the idea that mesopredators maintain Sarcocystis life cycles, as of species that infect roe deer (Entzeroth, 1981).
Another explanation for the effect of wolves on parasites in red deer and the apparent absence of a similarly strong effect in roe deer could be that both cervids differ in their regional distribution.
Recolonizing wolves excreting Sarcocystis oocysts with their feces could now be bridging a rather patchy distribution of roe and red deer, with little overlap in terms of co-occurrence in the same habitat in eastern Germany. There are currently no data on the distribution patterns of these ungulates from the two study regions in Germany, so it is unclear whether this is actually the case. Personal observations on hunting bags showed that usually one cervid species dominated the hunting bag when samples were collected in a particular area, consistent with findings from a study on deer densities in Scottish forests (Latham, Staines, & Gorman, 1997). Here, high red deer densities had a significant negative influence on roe deer densities. A recent study by Wu and colleagues showed that suitable habitats for red and roe deer do not necessarily overlap (Wu, Li, & Hu, 2016), even though it is generally accepted that they are sympatric species.

| Comparison of Sarcocystis communities between intermediate (ungulate) and definitive (wolf) hosts
In

| Age-related Sarcocystis infection in wolves
Experimental approaches in Toxoplasma serve as a model to understand the immunological response of hosts in their interaction with parasites in protozoan infections (Leng, Butcher, & Denkers, 2009 Tonelli, 2016) or, in the well-studied case of Toxoplasma, feline species (Rush, Lappin, & Milhausen, 2001).
Even though it is not clear which molecular mechanisms are responsible for Sarcocystis defense in wolves, wolf pups hosted more Sarcocystis species than older animals, consistent with the predictions from the "fading out" hypothesis that immunological resistance to Sarcocystis is higher in adults. This finding is also consistent with studies of domestic dogs where Cystoisospora and Giardia infections were most prevalent in pups (Barutzki & Schaper, 2003;Bugg, Robertson, Elliot, & Thompson, 1999). In wild canids, comparable indications of age-related parasite burden have previously only been reported in helminth etiopathology (Guberti, Stancampiano, & Francisci, 1993;Veronesi et al., 2014;Webster et al., 2017). In this study, we investigated this phenomenon accepting that coevolution in a host-parasite arms race can drive hosts to counteract parasite invasion by developing immune defense mechanisms. In turn, parasites are expected to evolve strategies to circumvent such barriers (Dawkins & Krebs, 1979;Schmid-Hempel, 2011), which can result in a higher persistence of such parasites despite an increasing immune competence of maturing host individuals. Sarcocystis grueneri and S. taeniata appear to be particularly well adapted to wolves as indicated by their lower decrease in prevalence during wolf maturation compared to other Sarcocystis spp. If these parasites have adapted to wolves, it may represent a more subtle adaptation than immune escape and might benefit from further investigation.
In wild boar, one of two known Sarcocystis species was identified.
We confirmed S. miescheriana (Kia, Mirhendi, Rezaeian, Zahabiun, & Sharbatkhori, 2011;Prakas & Butkauskas, 2012) but not the zoonotic S. suihominis (Prakas & Butkauskas, 2012) in our sample of wild boar ( Figure S2c). The fact that the supposedly suid-specific species S. miescheriana was also detected in cervids, and that S. miescheriana reads sequenced in this study had a higher intraspecific diversity than all other detected sarcocystis (Table S1), suggests that S. miescheriana sequences deposited in GenBank could potentially derive from more than the one species characterized to date.