Parasite infections in a social carnivore: Evidence of their fitness consequences and factors modulating infection load

Abstract There are substantial individual differences in parasite composition and infection load in wildlife populations. Few studies have investigated the factors shaping this heterogeneity in large wild mammals or the impact of parasite infections on Darwinian fitness, particularly in juveniles. A host's parasite composition and infection load can be shaped by factors that determine contact with infective parasite stages and those that determine the host's resistance to infection, such as abiotic and social environmental factors, and age. Host–parasite interactions and synergies between coinfecting parasites may also be important. We test predictions derived from these different processes to investigate factors shaping infection loads (fecal egg/oocyte load) of two energetically costly gastrointestinal parasites: the hookworm Ancylostoma and the intracellular Cystoisospora, in juvenile spotted hyenas (Crocuta crocuta) in the Serengeti National Park, in Tanzania. We also assess whether parasite infections curtail survival to adulthood and longevity. Ancylostoma and Cystoisospora infection loads declined as the number of adult clan members increased, a result consistent with an encounter‐reduction effect whereby adults reduced encounters between juveniles and infective larvae, but were not affected by the number of juveniles in a clan. Infection loads decreased with age, possibly because active immune responses to infection improved with age. Differences in parasite load between clans possibly indicate variation in abiotic environmental factors between clan den sites. The survival of juveniles (<365 days old) to adulthood decreased with Ancylostoma load, increased with age, and was modulated by maternal social status. High‐ranking individuals with low Ancylostoma loads had a higher survivorship during the first 4 years of life than high‐ranking individuals with high Ancylostoma loads. These findings suggest that high infection loads with energetically costly parasites such as hookworms during early life can have negative fitness consequences.

The ability of an individual to prevent, control, and clear infection is termed resistance, whereas tolerance is defined as the ability to limit the damage caused to an individual's health status and the associated Darwinian fitness cost of a given infection load (Råberg, Graham, & Read, 2009). There is growing interest in the effect of GI parasite infections on health status and components of fitness such as survival, longevity, and reproductive success in mammals (Hayward et al., 2014;Kutz, Hoberg, Nagy, Polley, & Elkin, 2004;Lynsdale, Mumby, Hayward, Mar, & Lummaa, 2017) and how phenotypic expression of both resistance and tolerance changes during an individual's life span. Most studies focus on adults, even though juveniles may be more susceptible to infection and may suffer more severe outcomes of infection than adults (Chilvers, Duignan, Robertson, Castinel, & Wilkinson, 2009;East et al., 2008;Marcus, Higgins, & Gray, 2014;Shrestha et al., 2015).
Resistance to parasite infection depends on both intrinsic and extrinsic factors (Ardia, Parmentier, & Vogel, 2011;Hayward et al., 2011;Jackson et al., 2011). In mammals, age is one intrinsic factor that can alter resistance to infection, because immune processes generally improve from early life to adulthood and then decline in old age (Simon, Hollander, & McMichael, 2015). When young, the cellular immune responses of juveniles typically differ both qualitatively and quantitatively from those of adults (Ramsburg, Tigelaar, Craft, & Hayday, 2003;Watson et al., 2016) and juveniles are less likely to have the acquired immunity that develops following exposure to pathogen antigen than adults, (Cattadori, Boag, Bjørnstad, Cornell, & Hudson, 2005;Ferreira, Torelli, et al., 2019), and thus, juveniles are generally more vulnerable to infection than adults.
The period of rapid growth and development during the juvenile life-stage is energetically costly, and when food intake is insufficient, juveniles may resort to resource allocation trade-offs to support growth and development at the expense of immune responses (Roff, 1992;Sheldon & Verhulst, 1996;Stearns, 1992). When this occurs, resistance to GI parasite infection should decline and infection loads increase. Furthermore, high infection loads of energetic costly parasites might compromise a host's nutritional status, thereby curtailing resource allocation to immune processes. When this leads to an increase in host susceptibility to infection, the outcome may be increased infection loads and an increase in the number of coinfecting taxa. This negative cycle might have detrimental fitness consequences such as reduced survival Beldomenico et al., 2008). Reduced survival in individuals with high infection loads indicates a lower tolerance of infection than that in animals that survive these levels of infection.
An individual's behavior may determine its chance of pathogen infection and the outcome of infection. For example, the aggregation of animals at breeding sites and water sources is thought to promote both the direct transmission of pathogens between hosts and the level of environmental contamination of such sites with infective stages (Cattadori et al., 2005;East, Kurze, Wilhelm, Benhaiem, & Hofer, 2013;Stommel, Hofer, Grobbel, & East, 2016). In social mammals, interactions between members of a group can affect the chance and outcome of infection for an individual (Duboscq, Romano, Sueur, & MacIntosh, 2016;Kappeler, Cremer, & Nunn, 2015;Nunn, Jordan, McCabe, Verdolin, & Fewell, 2015). Within-group social status (social rank) emerges from the outcome of behavioral interactions between animals in a group, and typically, high-ranking animals are more attractive social partners than low-ranking animals (Seyfarth, 1977). This suggests that more frequent social interactions by highranking individuals enhance the likelihood of contact with pathogens and hence they may require greater resistance against infection and/ or a higher tolerance of higher infection loads than low-ranking individuals (Marescot et al., 2018). High-ranking individuals in mammalian groups typically have greater access to food resources and a higher nutritional status than low-ranking individuals, and as a result, offspring reared by high-ranking mothers are often better nourished than those reared by low-ranking mothers (Clutton-Brock & Huchard, 2013;Hofer, Benhaiem, Golla, & East, 2016). This suggests that juveniles reared by high-ranking mothers (high-ranking juveniles) should find it easier to allocate the energy and protein (Jones et al., 2011) required for the maintenance of effective immune processes and the repair of parasite damaged tissue than those reared by low-ranking mother (low-ranking juveniles). As immune processes and tissue repair are key components of resistance to infection, highranking juveniles should experience a less severe impact of infection (for any given infection load) on their health and on fitness components such as survival than low-ranking juveniles. Thus, high-ranking juveniles are expected to exhibit greater tolerance of infection than low-ranking juveniles. Several studies provide evidence that adult high-ranking individuals do allocate more resources to immune processes and the repair of parasite damaged tissue than low-ranking animals (Ardia et al., 2011;East et al., 2015;Flies, Mansfield, Flies, Grant, & Holekamp, 2016;Marescot et al., 2018), but studies of this kind on juveniles are rare.
A range of intestinal parasites infect wild spotted hyenas (Crocuta crocuta) (Baylis, 1937;Berentsen et al., 2012;East et al., 2013;Engh et al., 2003;Graber & Blanc, 1979), and recently, a metabarcoding study extended knowledge on the composition and diversity of the bacterial microbiome and eukaryome in this species (Heitlinger et al., 2017). Spotted hyenas (hereafter termed hyenas) are social carnivores that live in multifemale, multimale fission-fusion societies termed clans (Kruuk, 1972) that defend territories (Frank, 1986). In the Serengeti National Park (NP), the main prey species are migratory ungulates, and as a result of their movements, there are large fluctuations in food abundance in clan territories (Hofer & East, 1993a, 1993b. High-ranking females have priority of access to food resources in their clan's territory and thus forage for a substantial proportion of each year in their territory. In contrast, for a substantial proportion of each year, low-ranking females travel long distances (from 80 to 140 km per foraging trip) to forage outside their territory in areas containing large herds of migratory ungulates (Hofer & East, 1993b). The greater the distance females travel the larger the number of days between nursing visits to their dependent offspring stationed at communal dens in clan territories (Hofer & East, 1993c).
Low-ranking mothers have longer internursing intervals , and their offspring have lower growth rates and are less likely to survive to adulthood than offspring of high-ranking mothers . Females reproduce throughout the year, and their offspring are dependent on milk for the first 6 months of life and are weaned between 12 and 18 months of age Hofer & East, 1993c).
Our study aims to investigate determinants of both Ancylostoma (nematode) and Cystoisospora (coccidian) infection loads in high-ranking and low-ranking juvenile spotted hyenas and to assess the resistance and tolerance of juveniles in these two social categories to parasite infection. We focus on Ancylostoma and Cystoisospora because both have direct life cycles that need no intermediary host, both cause damage to the epithelial lining of the small intestine, and are considered energetically costly parasites (East et al., 2015;Seguel & Gottdenker, 2017;Shrestha et al., 2015). We test predictions derived from six hypotheses: (a) the resource allocation hypothesis of life-history theory expects high-ranking juveniles to have lower infection loads than low-ranking juveniles, because they should have more resources to allocate to immune processes. (b) An improvement of juvenile immune responses with age should result in a decline in infection loads with age. Higher resistance is thus expected in high-ranking than low-ranking juveniles and in older than younger juveniles. (c) The transmission of Cystoisospora by the fecal-oral route during social contacts predicts higher infection loads in high-ranking than low-ranking juveniles, if resistance to infection is similar in these rank categories. Finally, we expect the survival of young juvenile hyenas to adulthood and their longevity to decrease with increasing Ancylostoma and/or Cystoisospora infection loads, and the fitness cost of high infection loads to be less in high-ranking juveniles than low-ranking juveniles, that is, we expect high-ranking juveniles to be more tolerant of high infection loads.

| Study host population
The study was conducted between March 2010 and August 2011 in three large clans that held territories at the junction between the woodland and plains in the center of the Serengeti NP. As this area is not within either the wet or dry season ranges of the migratory ungulates, females in all three clans conducted regular long distance foraging trips to locations outside their clan territory (East, Burke, Wilhelm, Greig, & Hofer, 2003;Hofer & East, 1993a).
The mean number of animals (±SEM, plus minimum and maximum values) in these clans during the study period was 79.4 ± 0.6 adults (range 67-86 adults), 22.5 ± 0.6 juveniles (range 11-32 juveniles) in the Isiaka clan, 66.0 ± 0.6 adults (range 54-69 adults), 28.8 ± 0.4 juveniles (range 22-32 juveniles) in the Pool clan, and 81.5 ± 0.9 adults (range 67-86 adults), 20.0 ± 0.7 juveniles (range 12-28 juveniles) in the Mamba clan, respectively. These clans have been studied since May 1987, November 1989, and August 1990, respectively (East et al., 2003. Information on individual longevity used in this study was collected until the end of July 2018. Individuals were recognized by their unique spot patterns and other features (Frank, 1986;Hofer & East, 1993a). Juveniles were aged using a range of characteristics (pelage, size, locomotion) when first detected, typically within their first few weeks of life, as previously described Golla, Hofer, & East, 1999). Sex was assessed at approximately 3 months of age using the dimorphic glans morphology of the erect clitoris or penis (Frank, Glickman, & Licht, 1991). Weaning typically occurs at 12-18 months of age (Hofer & East, 1995). Juveniles remain at the clan's communal den until approximately 12 months old where they shelter in underground burrows small enough to prevent entry of adults (Golla et al., 1999;Hofer & East, 1993c).
Juveniles obtain their social position as a result of the behavioral support they receive from their mother during interactions with other members of the clan (East et al., 2009;Smale, Frank, & Holekamp, 1993). As a result, juveniles were allocated the social rank held by their mother on the day the juvenile was sampled . To construct linear dominance hierarchies for females in each clan, we recorded ad libitum submissive behaviors during dyadic interactions between adult females during frequent observation periods of approximately three hours at dawn and dusk, mostly at the clan's communal den, as previously detailed . Dominance hierarchies were adjusted after each loss or recruitment of adults and when dyadic interaction data revealed that an individual had increased or fallen in rank. The social ranks held by the mothers of juveniles in this study changed little between March 2010 and August 2011. To permit the comparison of ranks held by individuals in different clans, we computed a standardized rank. This measure places the ranks within a given hierarchy evenly between the highest (standardized rank: +1) and the lowest (standardized rank: −1) rank Goymann et al., 2001). Juveniles were then categorized as high ranking if their mother's standardized rank was equal to or above 0, and low ranking if their mother's standardized rank was below 0 (as detailed by East et al., 2015;Marescot et al., 2018). We term juveniles reared by high-ranking mothers as highranking juveniles and juveniles reared by low-ranking mothers as low-ranking juveniles.

| Parasites
The GI parasite community in juveniles found by this study included helminths (Ancylostoma, Spirurida and Trichuris, Dipylidium, Diphyllobothrium, Taeniidae) and the apicomplexan Cystoisospora (Table 1). Infection with adult Ancylostoma hookworms is energetically costly (Seguel & Gottdenker, 2017) because adults live attached to the mucosal layer of the small intestine and feed on intestinal mucosa and blood. Adults shed eggs into the intestinal lumen which are voided in host feces (Sowemimo & Asaolu, 2008;Urquhart, Armour, Duncan, Dunn, & Jennings, 1996). Eggs hatch when environmental conditions (substrate, temperature and moisture) are suitable and larvae molt twice before they are infective. Infection occurs by ingestion or when larvae penetrate the skin of a host and then migrate through host tissue to the small intestines (Urquhart et al., 1996). Larvae can be transmitted to the offspring through the transmammary route during lactation in some host species (Burke & Roberson, 1985;Urquhart et al., 1996) or by ingestion of paratenic hosts with larvae infected tissue (Lee, Little, & Beaver, 1975). The pathogenic effects result from the blood loss instigated by adult worms and the damage adult worms cause to the intestinal epithelium (Urquhart et al., 1996).
Cystoisospora (formerly termed Isospora) infects epithelial cells lining the host's small intestine and is predominantly transmitted by the fecal-oral route (Lindsay, Dubey, & Blagnurn, 1997;Shrestha et al., 2015). Oocysts in feces can remain infective for weeks under favorable environmental conditions, and sporulation may take <16 hr at 30°C (Lindsay et al., 1997). Following the infection of intestinal epithelial cells, asexual reproduction occurs, thereby enhancing the infection load in these cells in relation to the infection dose. During this phase, histological changes in the small intestine occur (Lindsay et al., 1997;Shrestha et al., 2015) resulting in clinical signs such as diarrhea and weight loss (Lindsay et al., 1997;Mengel et al., 2012).
Sexual reproduction occurs after a period of several to many days, depending on the species of Cystoisospora. Oocysts are then shed in the host's feces (Lindsay et al., 1997;Mengel et al., 2012). For details on life cycles of spotted hyena GI parasites other than Ancylostoma and Cystoisospora, please see Table S1.

| Parasite screening
Fresh feces (N = 154) were immediately collected after deposition from 96 known juveniles (N = 50 males; N = 46 females) between 36 and 726 days of age (median age 188 day, mean age 214 days). Each fecal sample was mechanically mixed, and one aliquot was stored in a 4% formalin solution at room temperature until processed.
We determined fecal egg counts (FECs, eggs/g feces) and fecal oocyst counts (FOCs, oocysts/g feces) as previously described (East et al., 2015;Heitlinger et al., 2017) using the McMaster egg flotation technique following Foreyt (2001). We screened 2 g of feces per sample and used a potassium iodide (KI) solution with a specific weight of 1.5 g/ml and with a dilution ratio of 1:15 (Meyer-Lucht & Sommer, 2005). The McMaster slide counting chambers were loaded and left for 5 min before parasite eggs/oocysts were identified, measured, and counted in four chambers per sample using a compound microscope at 100× magnification (Jenaval, Carl Zeiss). Taxa were identified by conventional morphological criteria using veterinary parasitology reference manuals (Bowman, Hendrix, & Lindsay, 2002;Eckert, Friedhoff, Zahner, & Deplazes, 2008;Foreyt, 2001;Zajac & Conboy, 2012). Taxonomic identification to the resolution of order, family, or genus level varied between taxa. Results are presented as eggs or oocysts per gram of feces (EPG or OPG). These values were calculated by dividing the total number of eggs counted by the mass of feces in the counted chambers, which corresponds to the mass of feces measured per sample (2 g) divided by the volume of KI solution used (28 ml) and Note: Indicated are the percentage of infected host individuals (prevalence, %), mean and median infection loads of infected hosts (mean and median intensity, number of eggs or oocysts/g feces), mean infection load across all hosts, including the noninfected hosts with a load of 0 (mean abundance, number of eggs or oocysts/g feces) and the ratio of the variance to the mean of abundance. multiplied by the total volume counted (0.6 ml). To investigate the reliability of detecting the presence or absence of infection at low parasite loads, we repeated the analysis using three samples with low parasite egg loads. This revealed that the chance of detecting Dipylidium was 33% (2 of 6 counts), Trichuris 50% (7/14), Taeniidae 83% (5/6), Spirurida 100% (4/4), and Cystoisospora 100% (4/4). In three of five taxa, there was a chance of a false-negative result; hence, results from these three taxa should be interpreted with caution.
For Ancylostoma duodenale, fecundity (the number of eggs/g feces per worm) is density-dependent, declining from 287 eggs to approximately 100 eggs for the first 100 worms and then remains largely independent thereafter (Anderson & Schad, 1985); hence,

| Data analysis
The prevalence of infection is the proportion of individuals infected by a particular parasite taxon. Infection load corresponds to the FEC or FOC of a particular parasite taxon per feces. Mean intensity of infection describes the mean value of infection load of a particular parasite taxon among infected hosts. Mean abundance describes the mean of infection load of a particular parasite taxon among all host examined (Bush, Lafferty, Lotz, & Shostak, 1997;Margolis, Esch, Holmes, Kuris, & Schad, 1982). The ratio of the variance to the mean abundance was calculated to assess the degree of overdispersion (Table 1).
We used infection loads for Ancylostoma and Cystoisospora and the presence or absence of coinfection with the other parasite taxa identified in our study (Table 1) as recommended by Fenton, Viney, and Lello (2010). We applied generalized mixed-effect linear models (GLMM) using the package "lme4" (Bates et al., 2015) in which Ancylostoma load (Table 2) and transformed log 10 (1+) Cystoisospora load (Table 3) were the response variables. Because the raw data were overdispersed, we used a negative binomial distribution with a log link function (Hilbe, 2011) and confirmed that negative binomial regressions adequately accounted for overdispersion. We used individual ID as random effect to prevent pseudoreplication of 38 individuals contributing more than one fecal sample.   We used log-likelihood ratio tests and the Akaike information criterion (AIC) in R (R Core Team, 2015) to check whether the complete models were superior to reduced models. Models were considered similar if differences in AIC were <2.5 and preferable if the difference exceeded 6.0 (Hilbe, 2009). The significance of each predictor variable was assessed as the marginal contribution of each parameter to the full model by subtracting from the full model the log-likelihood ratio of a second model with each specific parameter removed and testing the difference against a chi-square distribution with appropriate degrees of freedom (see discussions in Hilbe, 2011;Hosmer, Lemeshow, & Sturdivant, 2013). To assess the global goodness of fit, we used log-likelihood ratio tests to compare full models with intercept-only models. In preliminary models, we included season in terms of wet and dry season and sex as fixed effects but neither had a significant effect nor improved the models as described above.
To investigate factors influencing survival to adulthood at 2 years of age, we selected samples from juveniles younger than 12 months (364 days To investigate the tolerance of juveniles to high Ancylostoma infection load and to infection with Cystoisospora, we determined the longevity (in days) of those juveniles sampled before the age of 6 months (180 days). We applied this cutoff because juveniles <180 days old had the highest Ancylostoma and Cystoisospora infection loads (see results), were entirely dependent on milk, and hence, their intake of milk was likely to be sensitive to maternal rank-related access to food and the period in days between maternal nursing visits . Longevity was scored as an exact value if the death date was known or as a right-censored value (a) at the age of dispersal for males that dispersed (and whose fate could not be subsequently followed), (b) at their age on the day of the road accident for individuals killed in road accidents, and ( This survival analysis was conducted using the R package survival v.2.38 (Therneau, 2015).
In all statistical models, the significance threshold was fixed at 5% and all tests were two-tailed. All statistical analyses were performed using R version 3.2.2 (R Core Team, 2015). Unless otherwise indicated, results are presented as means ± SD

| RE SULTS
We quantified eggs or oocysts from seven parasite taxa in the feces of juvenile hyenas (Table 1). Infection prevalence was 94.2% for F I G U R E 1 Survivorship of juvenile spotted hyenas sampled before 6 months (180 days) of age (n = 50). (a) Juveniles with either high (HL) or low (LL) Ancylostoma egg load and either a high-ranking (HR) or low-ranking (LR) mother; (b) juveniles either infected (positive) or not infected (negative) with Cystoisospora and either a high-ranking (HR) or low-ranking (LR) mother. Rightcensored measures of survival are marked as tick marks and include males that dispersed (and whose fate could not be subsequently followed), animals killed by road accidents and females that survived until the last day of observation  and Diphyllobothrium in Figure 3a-c, respectively, for all individuals sampled on at least three dates (N = 46 samples from 13 individuals). In general, infection load declined with age for these three most prevalent parasite genera.
Concurrent infection with more than one parasite taxon was frequent. Individuals were infected with 0-5 parasite taxa, the mean number of parasite taxa per individual was 2.8 ± 1.2 (N = 154). The frequency distribution of the number of different parasite taxa found in individual hyenas is illustrated in Figure 4.
Survival to adulthood among juveniles sampled in the first 12 months of life (log-likelihood ratio test, G = 33.38, df = 6, p < 0.001) decreased as Ancylostoma load increased but was independent of Cystoisospora load (Table 4). The effect of maternal social status was modulated by age. In low-ranking juveniles, survival early in life was substantially lower than that of high-ranking juveniles. As age increased, survival to adulthood of low-ranking juveniles increased steeply, whereas survival to adulthood of highranking juveniles increased modestly with age ( Figure 6). Kaplan-Meier survivorship curves indicated that during approximately the first 4 years of life, high-ranking juveniles with low Ancylostoma loads ( Figure 1a) had an overall better survivorship than those with high Ancylostoma loads. The survivorship of low-ranking juveniles with high Ancylostoma loads was mostly below that of low-ranking juveniles with low Ancylostoma loads during approximately the first 5 years of life (Figure 1a). In contrast, high-ranking juveniles infected with Cystoisospora had a higher survivorship until early adulthood than those that were not infected with this parasite, whereas low-ranking juveniles that were infected with Cystoisospora had a lower survivorship than those not infected with Cystoisospora during approximately their first 5 years of life ( Figure 1b).

| D ISCUSS I ON
In line with other studies on parasite infections in free-ranging wild mammals, our study on spotted hyenas revealed substantial heterogeneity between juveniles in both their Ancylostoma and Cystoisospora infection loads ( Figure 2) and their number of coinfecting parasite taxa (Table 1, Figure 4). Consistent with evidence that immune processes in juvenile mammals improve with age, both Ancylostoma (Table 2) and Cystoisospora (Table 3) loads decreased as juveniles increased in age. We interpret differences in infection loads between juveniles in different clans as possible evidence that abiotic conditions at communal dens sites that influence the survival of parasite infective stages differed between clans. In contrast to our prediction, Ancylostoma and Cystoisospora loads decreased with the number of adults in a clan (Table 2). One plausible explanation could be an encounter-reduction effect whereby adult hyenas reduce the chance of juveniles encountering Ancylostoma infective larvae. Also contrary to expectation, we found no evidence that Cystoisospora loads were affected by social interactions among juveniles (Table 3) (Table 4) decreased as Ancylostoma infection loads increased but was not affected by Cystoisospora infection loads (Table 4). Furthermore, the longevity of young juveniles with high

Ancylostoma infection loads was poor compared to those with low
Ancylostoma infection load, and this was particularly so in juveniles reared by low-ranking mothers (Figure 1). These findings indicate that energetically costly parasites such as Ancylostoma can have negative fitness consequences for juveniles and that high-ranking juveniles may be more tolerant of infection (i.e., have lower fitness costs) than low-ranking juveniles.
Juvenile hyenas in our study population are entirely dependent on maternal milk for their growth during their first 6 months of life .  (Table 4) and that high-ranking juveniles had a higher probability of reaching adulthood than low-ranking juveniles (Table 4, Figure 6). Furthermore, survival to adulthood increased with age and there was an interaction between juvenile age and maternal rank (Table 4). During the first 6 months of life, the probability of survival for low-ranking juveniles was very low and for the few low-ranking juveniles that did survive this initial period, their probability of survival was then similar to the higher survival of high-ranking juveniles ( Figure 6). We also found evidence that in both high-ranking and low-ranking juveniles, survivorship (longevity) appears to be less in animals with high than those with low Ancylostoma loads. High-ranking juveniles with high infection loads lived longer than low-ranking juveniles with high loads (Figure 1a).
This result is based on a relatively small number of juveniles, and thus, the effect of Ancylostoma loads on longevity should be further investigated with a larger sample of juveniles. Our results indicate that most animals in our study were infected with Ancylostoma.
Given this high prevalence of Ancylostoma infection (Table 1) and the absence of a maternal rank effect on Ancylostoma loads (Table 2), we interpret these results to indicate that high Ancylostoma loads decrease longevity. Furthermore, we interpret the lack of a maternal rank effect on Ancylostoma loads (Table 2) to suggest that higher milk intake and higher growth rates in high-ranking than low-ranking juveniles  do not provide evidence that high-ranking juveniles are more resistant to infection than low-ranking juveniles (Figure 1a), but rather that better nourished high-ranking juveniles have a higher "tolerance" phenotype to Ancylostoma load than low-ranking juveniles. This idea needs to be more rigorously tested with a larger sample of juveniles, as it is possible that, in combination with other factors, even low Ancylostoma loads may reduce longevity.
Previously, we have shown that offspring reared by low-ranking mothers have lower growth rates than those reared by high-ranking females . This is because low-ranking mothers in our study population more often travel long distances (round trips of up to 140 km) to forage outside the clan territory than high-ranking mothers (Hofer & East, 1993c. As a result, during the first 6 months of life, when growth is fueled by milk, low-ranking juveniles are nursed less often and overall receive less milk than highranking juveniles . Furthermore, juveniles in our study population with low growth rates in the first 6 months of life have a lower probability of survival to adulthood than those with high growth rates . High Ancylostoma infection loads are energetically costly because of substantial blood loss and damage to the intestinal wall (Urquhart et al., 1996), and have been reported to retard juvenile growth, and cause anemia and malnutrition in other mammals (Sakti et al., 1999;Seguel & Gottdenker, 2017;Stoltzfus et al., 2004). This suggests that the energetic cost of high Ancylostoma infection loads in rapidly growing, milk-dependent juvenile hyenas may compromise growth and thus curtail survival to adulthood. This idea is supported by evidence that pup growth in New Zealand sea lion (Phocarctos hookeri) is reduced by infection with the hookworm Uncinaria sp (Chilvers et al., 2009), and body mass, fat deposits, and fecundity in reindeer (Rangifer tarandus platyrhynchus) are reduced by the nematode Ostertagia gruehneri (Stien et al., 2002). Using all these strands of evidence, we interpret our current finding to suggest that high-ranking juveniles are more tolerant of high Ancylostoma infection loads than low-ranking juveniles, because they are generally better nourished and thus can allocate more resources to replace the blood lost to feeding adult Ancylostoma and to repair the damage to the intestinal wall than lowranking juveniles. Similarly, a study of nematode infections in female Soay sheep on the island of St Kilda (Hayward et al., 2014) reported considerable variation in weight loss in relation to a given strongyle egg load and that animals that lost weight more slowly had a higher lifetime breeding success, indicative of a greater tolerance of infection and higher fitness. Tolerance is a trait that is thought to lead to a higher infection prevalence of a parasite in a host population and this should promote the persistence of infection (Roy & Kirchner, 2000).
Consistent with this idea, our results from juvenile hyenas and those of a previous study (East et al., 2015) indicate both a high prevalence and persistence of Ancylostoma infection in our study clans.
Although we found no evidence that high Cystoisospora infection loads decreased the survival of juvenile hyenas to adulthood (Table 4), the survival probability for low-ranking individuals that were infected with Cystoisospora when they were juveniles was lower throughout the juvenile period and into early adulthood than that of low-ranking juveniles not infected by Cystoisospora (Figure 1b).
Our study does not assess infection loads during the first few weeks after birth when we are unable to obtain feces from cubs; hence, we cannot rule out the possibility that infection during this period might decrease survival. Infection with Cystoisospora suis within days of birth of domestic piglets can cause severe pathologies, increased mortality and poor weight gain (Koudela & Kučerová, 1999;Lindsay et al., 1997;Shrestha et al., 2015), and an enhanced chance of coinfection (Mengel et al., 2012). Furthermore, our analysis does not distinguish between juvenile hyenas with chronic Cystoisospora infections that may curtail juvenile survival to adulthood from those with relatively short-term, high infection loads that may have a limited effect on juvenile survival (Figure 3b). A preponderance of such short-term infections in our sample may contribute to the lack of a significant effect of high oocyst infection load on juvenile survival to adulthood.
Our analysis found no evidence for the expected positive relationship between Cystoisospora loads in juveniles and the number of juveniles in a clan (Table 3). This relationship was expected because Cystoisospora is transmitted by the fecal-oral route and social interactions, particularly greeting ceremonies between young juveniles, in which participants sniff and lick their partner's anal genital region (East, Hofer, & Wickler, 1993), might be expected to increase with the number juveniles in a clan (Olarte-Castillo et al., 2016). Environmental contamination of communal den sites might also be expected to increase with the number of juveniles, thereby raising the chance of inadvertent ingestion of oocysts by juveniles.
We studied three clans that contained roughly a similar mean number of juveniles (between 20 and 29 juveniles). It is possible that the inclusion of a larger number of clans that differed substantially in the number of juveniles they contained might yield results that differ from ours.
Our results revealed that heterogeneity in Cystoisospora oocyst loads shed by juveniles was considerable (Figures 2, 3b) and that high infection loads occurred mostly early in life ( Figure 3b) and declined as juvenile age increases (Table 3). These findings suggest that at any given time, high infection loads are probably mostly apparent in a few younger juveniles in a clan. It is possible that high infection loads in these individuals may be linked to rapid asexual reproduction of the parasite within the intestinal epithelial cells of some individuals, followed by sexual reproduction and the shedding of high oocyst loads-rather than high infection loads resulting from are thought to be important for disease transmission in several pathogens (Courtenay, Peters, Rogers, & Bern, 2017;Lloyd-Smith, Schreiber, Kopp, & Getz, 2005;Paull et al., 2011). Because adult females in our study clans give birth throughout the year Hofer & East, 1995), for a substantial proportion of each year, communal dens are likely to contain the young juveniles ( Figure 3b) that typically shed high oocyst loads. Communal dens are likely to be hot spots for Cystoisospora transmission to susceptible individuals because both juveniles (this study) and lactating females at dens (East et al., 2015) shed oocysts in feces. Hyena communal den sites are locations of high use by both adults and juveniles ( Figure 7) and thus can act as hot spots for the spread of pathogens, similar to bird feeders (Adelman, Moyers, Farine, & Hawley, 2015) and savannah waterholes (Stommel et al., 2016). Both super-spreaders and hot spots of infection are important for the spread and maintenance of pathogens in host populations (Adelman et al., 2015;Courtenay et al., 2017;Lloyd-Smith et al., 2005;Paull et al., 2011). Juvenile hyenas with high infection loads at communal dens may be important for the spread and maintenance of not only Cystoisospora and Ancylostoma (this study) but other parasite such as Dipylidium (East et al., 2013).
For approximately 2 months after birth, defecation by young juveniles is stimulated by maternal grooming of their offspring's anus and mothers typically consume the feces produced (M. L. East and H. Hofer, personal observations). This maternal behavior may help limit oocyst contamination of dens.
Ancylostoma larvae emerge from eggs shed in feces and then molt twice before becoming infective filariform larvae that remain viable for weeks in moist and cool environmental conditions (Urquhart et al., 1996). Hyenas are infected when Ancylostoma infective larvae either burrow into their feet or pass from infected mothers to their offspring during nursing. We previously found that Ancylostoma infection loads were higher in low-ranking than high-ranking lactating females (East et al., 2015). These rank-related differences in maternal infection loads might be expected to result in the passage of more infective larvae from low-ranking than high-ranking mothers to their offspring during nursing. However, even though Ancylostoma egg load is a good indicator of the infection load of adult worms in a host (Anderson & Schad, 1985), the number of adult Ancylostoma worms in the intestine may not reflect the number of migrating larvae in the female's tissue or the number of larvae shed in her milk, and this may explain why there was no significant effect of maternal rank on Ancylostoma infection load in juvenile hyenas (Table 2). In several species of sea lions, transmission of the hookworm Uncinaria by the transmammary route is thought to be important (Marcus et al., 2014).
The three study clans varied in the mean number of adults they contained (between 66 and 82 adults), and although the number of clans studied was small, our results indicate that infection loads of Ancylostoma (Table 2) and Cystoisospora (Table 3) decreased as the number of adult clan members increased. This suggests an encounter-reduction effect (Côté & Poulin, 1995;Keesing, Holt, & Ostfeld, 2006;Mooring & Hart, 1992), which in multispecies host assemblages is termed a "dilution" effect. Evidence of an encounter-reduc-  (East et al., 2015) is typically less than that in feces produced by juveniles (this study), probably because of a more effective immune response to infection by adults than juveniles. Unless encountered in a fresh condition, the feces of adult hyenas are often extremely hard and dry (Kruuk, 1972), In line with the findings of previous studies that provide evidence that immunocompetence in juvenile mammals improves with age (Koudela & Kučerová, 1999;Ramsburg et al., 2003;Simon et al., 2015), both Ancylostoma and Cystoisospora infection loads in juvenile hyenas decreased with age (Tables 2, 3). This age effect was also apparent for infection with Ancylostoma, Cystoisospora, and Diphyllobothrium in juvenile hyenas sampled several times during their development (Figure 3a-c). The less developed immune responses of neonates develop during the juvenile period into the fully mature immune system of adults. For example, during early life, juveniles are generally more susceptible to infection by pathogens than adults, partly because of their limited capacity to mount an effective immune responses due to qualitative differences in the Tcell populations of juveniles and adults (Ramsburg et al., 2003), and naïve, mostly young animals that lack exposure to pathogen antigens have not developed the acquired immune responses that develop as a result of exposure (Cattadori et al., 2005;Koudela & Kučerová, 1999). The decrease in Ancylostoma and Cystoisospora infection loads in juvenile hyenas with age may be the combined outcome of increased immunocompetence due to active immune responses induced by parasite exposure, which is likely to increase with age Ferreira, Torelli, et al., 2019) and the general maturation of the mammalian immune system in juvenile hyenas with increasing age. Currently, little is known about factors that affect immunocompetence in juvenile hyenas, so in addition to age, factors such as genotype and gene expression (Gulland, Albon, Pemberton, Moorcroft, & Clutton-Brock, 1993;Jackson et al., 2011), allostatic load, physiological processes, behavior, and diet may be relevant (Ardia et al., 2011;VanderWaal & Ezenwa, 2016).
Our results revealed a significant effect of clan membership on both Ancylostoma (Table 2) and Cystoisospora (Table 3)  Note: Shown are the logistic regression coefficient estimates with their standard errors (SE) and 95% confidence limits in natural log units as well as their conversion into odds ratios with their respective 95% confidence limits (CI), the z-values and associated p-values for each parameter and the results of log-likelihood ratio tests (G) with associated p-values. Positive (negative) estimates indicate that an increase in the value of the parameter increased (reduced) the survival to adulthood. Survival for juveniles with low maternal rank was lower than for juveniles with high maternal rank. a Presence of one or more of the following taxa: Diphyllobothrium, Dipylidium, Taeniidae, Spirurida, Trichuris.
infection loads than those in the Pool clan. This finding suggests that the survival of eggs, oocysts, and infective larvae at den sites in the Pool clan territory was lower than those in the Mamba and Isiaka clan territories, possibly because of abiotic conditions that reduce survival, such as high ambient temperatures and low soil moisture (Turner, Versfeld, Kilian, & Getz, 2012;Urquhart et al., 1996). Clan differences in infection loads may also result from behavioral differences between juveniles in the Pool and other two clans. For example, juveniles in the Pool clan may have more often changed the location of the latrines they used than those in the other two clans, thereby avoiding latrines highly contaminated with infective stages. Similarly, the use of latrines in shorter vegetation where peak daytime temperatures were higher and soil moisture was lower, might also reduce the survival of infective stages and hence infection loads.
Most wild mammal populations are concurrently infected with more than one parasite (Behnke, Gilbert, Abu-Madi, & Lewis, 2005;Seltmann et al., 2019;Telfer et al., 2010), and this is also true for juvenile hyenas (Table 1, Figure 4). Our results revealed that as either Ancylostoma (Table 2, Figure 5a) or Cystoisospora (Table 3, Figure 5b) loads increased, the number of coinfecting GI parasite taxa hosted per juvenile also increased. The explanation for the positive relationship between infection loads and coinfecting taxa is unknown and requires further research. As maternal rank category did not have the predicted effect on either Ancylostoma (Table 2) or Cystoisospora infection loads (Table 3), we lack evidence that the immune processes of low-ranking juveniles are more often compromised by resource allocation trade-offs than those of highranking juveniles, and thus, the role of immune processes in determining the number of coinfecting parasite taxa is also unclear.
Even though potential interactions between taxa in the GI parasite community of juvenile hyenas are unknown, there was a tendency for an increase in juvenile survival to adulthood with the number of coinfecting GI parasite taxa per host (Table 4). This suggests the intriguing idea that the fitness cost of a more diverse GI parasite community may be lower than one dominated by a limited number of taxa. Previously, we found that the eukaryome in high-ranking hyenas is more diverse than that of low-ranking animals (Heitlinger et al., 2017) which may indicate that high-ranking animals have a more diverse, stable, and ecologically "healthy" GI community than low-ranking animals.
Our study identifies several factors that shape heterogeneity of parasite infections among juveniles in a wild social mammal.
Disentangling key environmental factors, parasite-host mechanisms, and interactions between coinfecting parasites is difficult, particularly when noninvasive methods are required, as is the case in many protected areas or in endangered species. The benefit of noninvasive methods is that they allow repeated measures of both infection loads and components of fitness throughout an individual's life span. Further research is needed to examine in detail the effect of the variation of clan size and composition on parasite infections.

ACK N OWLED G M ENTS
We are grateful to the Commission for Science and Technology of F I G U R E 6 Predicted effect of juvenile age and maternal social status on the probability of juvenile survival to adulthood for spotted hyenas younger than 12 months of age with 95% confidence intervals (n = 84). Alongside as "dots" are the raw data of juveniles included in the model that survived (probability of survival to adulthood = 1) and did not survive (probability of survival to adulthood = 0) to adulthood

DATA AVA I L A B I L I T Y S TAT E M E N T
The dataset analyzed for this study is available from the Dryad Digital Repository, https ://doi.org/10.5061/dryad.5qv7v47 (Ferreira, Hofer, Madeira de Carvalho, & East, 2019).