Parasites play an important role in the evolution of host traits via natural selection, coevolution and sexually selected ornaments used in mate choice. These evolutionary scenarios assume fitness costs for hosts. To test this assumption, we conducted an ectoparasite removal experiment in free-living Columbian ground squirrels (Urocittelus columbianus) in four populations over three years. Adult females were randomly chosen to be either experimentally treated with anti-parasite treatments (spot-on solution and flea powder, N = 61) or a sham treatment (control, N = 44). We expected that experimental females would show better body condition, increased reproductive success and enhanced survival. Contrary to our expectations, body mass was not significantly different between treatments at mating, birth of litter or weaning of young. Further, neither number nor size of young at weaning differed significantly between the two treatments. Survival to the next spring for adult females and juveniles was not significantly different between experimental and control treatments. Finally, annual fitness was not affected by the treatments. We concluded that females and their offspring were able compensate for the presence of ectoparasites, suggesting little or no fitness costs of parasites for females in the different colonies and during the years of our experiments.
Parasites and diseases play a major role in the evolution of organismal traits [1–3] through red queen coevolution, in which parasites and hosts coevolve in relation to one another . As a consequence, evolutionary host–parasite interactions occur. From the parasite's perspective, phenotypes that exploit the host most effectively have a higher individual fitness and should therefore be favoured by natural selection. On the other hand, it is advantageous for individual hosts to show enhanced parasite resistance, since their fitness depends on resources allocated to their own survival and reproduction. Another example of an evolutionary host–parasite interaction is the influence that parasite resistance can have on sexually selected ornaments (e.g. [5–7]), with lower attractiveness of infested animals to potential mates . Thus, resistance to parasites may have major evolutionary influences on host species and experimental studies are useful to evaluate the costs of parasitism in terms of fitness consequences.
Since natural selection operates on the diverse traits of host species across taxa (e.g. ), fitness consequences of parasites are expected owing to such influences as increased immune function  or the energy lost that cannot be used for other processes, including host reproduction, growth or survival . These fitness costs may be evident as slowed or lowered developmental rates (e.g. ), poor survival (e.g. ) or reduced reproductive rates (e.g. ). Such costs may be minimized over evolutionary time, since it may benefit both the parasite and host when the parasite is less virulent (reviewed in ). Most parasites do not kill, but have evolved to not harm the host beyond the point where the parasite's transmission is inhibited. To minimize energetic costs, hosts invest in immune-competence and parasite removal behaviour (e.g. scratching, grooming), which have to be balanced in fitness terms with reproduction, growth and survival [11,16]. The differences between parasitized and non-parasitized hosts could be owing to indirect effects, such as adaptations to specific parasites or host quality differences [17–19]. While an array of studies have been done in a laboratory environment, better understanding of the complexity of parasite–host interactions and the actual costs of parasites for hosts requires measurements of fitness trade-offs and impacts of parasites in experimental studies of wild animals . Even though evidence of host–parasite interactions under natural conditions is growing, much less is known regarding the cost of parasites in the wild (e.g. [21–23], but see e.g. [24–26]).
In an experimental study in free-ranging male Columbian ground squirrels (Urocitellus columbianus), we examined the ‘costs’ of parasitism in a group-living and sedentary species that might be expected to exhibit substantial parasite loads . Male ground squirrels were given an anti-parasite treatment that removed fleas and ticks, the major ectoparasites of ground squirrels (levels of endoparasites were lowered as well). This group was then compared to control males. The removal of parasites did not reveal costs for several fitness-related traits, such as behavioural mating rates and durations, mating position in the queue of mates for the polyandrous females, success at producing offspring per female or in total, and changes in body mass within or among seasons. These negative results were in contrast to the findings for female Columbian ground squirrels, in a study by Neuhaus , where a similar experimental ectoparasite removal experiment was conducted, and both body mass and numbers of offspring were significantly elevated by the ectoparasite removal. The different results on the impact of ectoparasites between the sexes were rather surprising, especially as male-biased parasitism is well documented in mammals [28–30]. This difference might be explained by the fact that Neuhaus' study  was conducted on a very small scale (four treated and five control females) and without measuring actual ectoparasites loads.
To help clarify these different results, we conducted another study with increased sample sizes and more parameters to confirm the fitness costs of parasites for free-ranging females in this species. Specifically, we re-examined the costs of ectoparasites for wild female Columbian ground squirrels, with an expanded sample from several ground squirrel colonies. To assess the actual cost of parasitism, we randomly applied either an experimental ectoparasite removal (and perhaps partial endoparasite removal ) to females and compared them to a sham treated control group. The fitness traits examined included mass (at the time of emergence from hibernation, mating, birth of the single annual litter of young, and at about the time when those young were weaned), litter size at weaning and survival of adult females and young to the next year. For litters of young that were infested with fleas at the time of weaning, we also examined litter size, maternal mass and offspring mass and survival. Finally, we investigated the impact of ectoparasites on an annual fitness measure that is based on maternal and offspring survival (after ).
Overall we predicted that experimental female ground squirrels would exhibit superior values of fitness measures compared to control females. If met, our findings would reveal the costs of ectoparasites for breeding females in terms of their somatic and reproductive efforts (sensu ) under natural conditions.
2. Material and methods
(a) Study species
From 2006 to 2008, we studied Columbian ground squirrels in the Rocky Mountains of southwestern Alberta, Canada, in the Sheep River Provincial Park (50° N, 114° W) on four different study sites that were separated by no more than 6 km (at elevations of about 1450–1550 m). Columbian ground squirrels are diurnal, colony-living rodents inhabiting subalpine and alpine meadows. Ectoparasites included flea species (identified in our colonies were Oropsylla opisocrostis tuberculata and Oropsylla oropsylla rupestris), mites (Dermacarus heptneri, Androlaelaps fahrenholzi, Macrocheles sp. and Pygmephorus erlangensis), lice (Enderleinellus suturalis and Neohaematopinus laeviusculus) and ticks (Ixodes sculptus) . In our four populations, adult males emerge first from hibernation in mid-April, followed by females a few days to a week later . Females breed on average four days after emergence from hibernation and are in oestrus for about 5–7 h during a single day . After 24 days of gestation females gave birth . The offspring subsequently emerge above ground, after 27 days of lactation, from special single-entrance ‘natal burrows’ at the time of weaning . The active season ends with six to seven weeks of fattening, in preparation for the eight- to nine-month hibernation season .
(b) Experimental procedure
Ground squirrels in the study meadows were captured shortly after emergence from hibernation (within two days) in live traps (13 × 13 × 40 cm, Tomahawk Live Trap Co.) baited with peanut butter. Individuals were weighed to the nearest 5 g with a spring-slide balance (Pesola Co.), permanently marked with numbered metal tags in both ears (Monel no. 1, National Band & Tag Co.) and marked on the dorsal pelage with a unique black dye mark (Lady Clairol human hair dye) for visual identification from elevated benches 3–5 m high. Condition of the female's vulva and visual observations of behaviour indicated the day that each female mated. Ground squirrels were observed daily during the mating period (about three weeks), and captured as necessary for treatments and at important stages in the life cycle (see below). At each capture, flea load was estimated (see below), individuals were again weighed and examined for reproductive condition and general health, and released at the point of capture.
In each colony, all reproductive females (two years or older) were randomly divided into two treatment groups: experimental or control. Details on the experimental parasite removal manipulation and sham treatment have been provided previously (for methodological details, see ). Briefly, roughly half of the reproductive females per colony were randomly chosen for treatment with a spot-on solution (Stronghold®; one drop per 100 g of body mass) and flea powder (Zodiac®; applied form a shaker and rubbed into the fur of the animal) to remove ectoparasites (n = 61 experimental females). The remaining females received a control, sham treatment by: (i) imitating the flea powder treatment by rubbing the fur without powder (massage) and (ii) applying a ‘spot’ of isopropyl-alcohol (alcohol used in the Stronghold® solution; n = 44 control females). Both spot-on and sham alcohol treatments were reapplied every 17 days, whereas the flea powder application or alternative massage were repeated every six days, until weaning. Ectoparasite load was estimated using a combination of combing and finger-stroking the fur and counting the number of visible fleas. A conscious effort was made to not remove fleas. Flea counts were used because fleas made up the vast majority of ectoparasites. For each captured ground squirrel, ectoparasite load was classed into four categories: (0) = no fleas detected, (1) = one to two fleas, (2) = three to five fleas, (3) = more than five fleas (range 6–15) . Ectoparasite load for each female was assessed at three time periods: directly after hibernation, before the treatment started; at mating; and at weaning of offspring.
(c) Mating and sampling of litters
Age at maturity varied from one to three years of age, but most females first bred at two years of age and continued to breed annually thereafter . Ages of females were not known with certainty on some of the meadows, but yearling, two-year-old and older females can be discerned by a combination of body mass (much lower for yearlings than older females at spring emergence) and condition of the nipples (folded under the skin for females that have not lactated before). After giving birth, mothers kept dependent young in single-entrance nest burrows during lactation . At about the time of weaning, juveniles were captured within 48 h of first emergence from nest burrows, in either National live traps or multiple-capture traps . The young were then ear tagged, weighed, dye marked and examined for fleas (0 = no ectoparasites, 1 = parasitized). The presence of fleas in underground nests at the time of weaning of young was inferred: in most such cases, all pups of a litter were heavily infested. These data provided an indirect estimate of nest infestation by fleas.
(d) Statistical analyses
To test for differences in ecotoparasite loads for experimental and control females before the experimental removal of ectoparasites (time period 1), a Kendell's τc was used. Subsequently, we applied an ordinal regression with treatment, time period and treatment × time period as fixed factors to examine the change of ectoparasite loads over time (periods 1–3) with respect to treatment (experimental and control). We examined only females that successfully weaned pups (total litter loss was independent of the type of treatment: χ2 = 1.5, d.f. = 1, p = 0.2) and for which complete ectoparasite load estimates over all time periods were available (control: n = 28; experimental: N = 35). We used SPSS 21 (SPSS Inc., Chicago, IL, USA) for analyses of ectoparasites loads.
Because different years and meadows were represented in our dataset, and because nine females appear in more than one year, we used mixed models to remove significant variation in dependent variables (e.g. litter size and survival) before testing for treatment effects. For this, we used mixed models to test for significant differences among years and meadows, holding treatment effects invariant as a random factor. We judged significance using a log-likelihood ratio test, where models were compared with and without year or meadow. When significant, years and/or meadows were included as random factors in a subsequent test using a mixed model to test for treatment effects. This proved to be a conservative procedure. Because some females were present in more than one year, individual identity was included as a random factor throughout. When testing for treatment effects on survival, we specified models that used a binomial distribution (and logit-link function) for the dependent variable. When testing for the effects of flea infestation of nests with young, treatment was included as a random factor to control for experimental effects.
Annual fitness was calculated as the survival of the mother (0 if not and 1 if she survived to the next spring) and 0.5 times the number of young that survived to the next spring as sub-adult yearlings (after ). Surviving offspring counted for only a half because they carry half their mother's genes. Spring emergence from hibernation when young were yearlings was the best time to estimate annual fitness, because males usually leave their natal areas when they are just over one year old [39–41].
We conducted conservative estimates of statistical power of the flea removal treatments compared to control females for reproduction, survival and annual fitness, using simple t-tests and exact binomial tests. These analyses were conservative because the more robust mixed models accounted for significant variations in the dataset owing to measurements in different years and meadows (see above). Estimates of power indicated the size of potential effects that would have been significant at the α = 0.05 level, given our actual sample sizes and variances of measured variables. We assumed that directional predications of the effect of flea removal treatments were appropriate, and thus used one-tailed tests.
Statistical tests of treatment effects and litter infestations by fleas were implemented using R v. 3.0.2 (R Foundation for Statistical Computing © 2013) in R Studio version 0.98.501 (RStudio, Incorporated, © 2009–2013). Mixed models used the glmer routine of the lme4 package.
(a) Ectoparasite load
Ectoparasite loads did not differ between experimental and control females before the experimental removal of ectoparasites (time period 1; Kendell's τc = –0.04, N = 63, p = 0.73; figure 1). Furthermore, flea load significantly decreased over time (from time period 1 and 2 to 3) in experimental females compared to control females (treatment: d.f. = 1, p < 0.001; time period 1: d.f. = 1, p < 0.001; time period 2: d.f. = 1, p < 0.001; treatment × time period 1: d.f. = 1, p < 0.001; treatment × time period 2: d.f. = 1, p < 0.001; time period 3 was used as the reference category; figure 1). A parallelism test revealed a different reaction over time for the two treatments (, p < 0.001), namely that the reduction of ectoparasites in experimental females was different from the changes found in control females. At the end of the experiment (time period 3), all 35 experimental females were ectoparasite free, whereas 71% of 28 control animals were still infested with ectoparasites.
(b) Body mass
Over the four study sites and over the course of three years, we treated 105 females (experimental = 61 and control = 44), out of which 78 mothers (experimental = 43 and control = 35) successfully weaned young. For changes in body mass during reproduction, years did not significantly differ, but meadows exhibited significantly different patterns (log-likelihood test for year effect: χ2 = 0.51, d.f. = 2, p = 0.77; log-likelihood test for meadow effect: χ2 = 21.14, d.f. = 3, p < 0.0001; respectively). Hence, we specified meadow and individual as random effects in a mixed model. Body mass changed significantly among the four time periods (figure 2; log-likelihood test for differences among time periods: χ2 = 324.60, d.f. = 3, p < 0.0001). Changes in body mass were not significantly different between experimental and control females (log-likelihood test for treatment effect: χ2 = 0.11, d.f. = 1, p = 0.74).
Offspring mass at weaning (controlled for litter size) did not significantly differ between two years, but differences among meadows were significant (log-likelihood test for year effect: χ2 = 2.62, d.f. = 2, p = 0.27; log-likelihood test for meadow effect: χ2 = 7.69, d.f. = 3, p = 0.05; respectively). Therefore, we included meadow and litter size as random factors in a mixed model (no mothers were repeatedly sampled for offspring mass). Offspring mass at weaning was not significantly different between treatments (log-likelihood test: χ2 = 0.78, d.f. = 1, p = 0.38). Mean offspring mass was 107.1 (±5.9 s.e., N = 33) for control mothers and 105.7 (±4.0 s.e., N = 43) for experimental mothers (t-test with variances assumed unequal, t = 0.19, d.f. = 58.1, p = 0.85). Offspring size and number at weaning were negatively correlated (r = −0.621, N = 76, p < 0.0001).
Among mothers, some had nest burrows and young at weaning that were highly infested with fleas. These provided a ‘natural’ comparison of high and low ectoparasite treatments, but only for females that weaned offspring. Infestation did not differ significantly among years or among meadows (log-likelihood test for year effect: χ2 = 0.92, d.f. = 2, p = 0.63; log-likelihood test for meadow effect: χ2 = 3.47, d.f. = 3, p = 0.32). Thus, we constructed a mixed model to ascertain the influence of the ectoparasite removal treatment on infestation of nests with young and included only female identity as a random factor. The treatment had a significant influence of whether nests of young were infested (log-likelihood test for year effect: χ2 = 35.40, d.f. = 1, p < 0.0001). Among natal nests of control mothers, 69.7% were infested (N = 33), while among natal nests of experimental mothers only 7.0% were infested (N = 43). Changes in body mass during reproduction were not significantly different between 26 infested and 50 un-infested mothers (log-likelihood test for infestation effect: χ2 = 0.49, d.f. = 1, p = 0.48). There was a trend towards heavier offspring at weaning for non-infested compared to infested natal nests (log-likelihood test: χ2 = 3.14, d.f. = 1, p = 0.08). Mean offspring mass was 109.7 (±4.05 s.e., N = 50) for non-infested nests and 99.7 (±6.05 s.e., N = 26) for infested nests (t-test, t = 1.37, d.f. = 47.6, p = 0.18).
(c) Reproduction success
There was no significant impact of year or meadow in the number of offspring at weaning (χ2 = 2.07, d.f. = 2, p = 0.36; χ2 = 3.66, d.f. = 3, p = 0.30; respectively). So we specified a mixed model with female identity as the only random factor. Mean litter size at weaning was very similar between control and experimental (removal) females (table 1a; log-likelihood test for treatment effect: χ2 = 0.01, d.f. = 1, p = 0.94). We also compared nests that were and were not infested with fleas, and found that non-infested nests had similar litter sizes compared to those in infested nests (table 1a; χ2 = 0.21, d.f. = 1, p = 0.65).
The survival of young to yearling age was significantly different between two years and three meadows (χ2 = 3.69, d.f. = 1, p = 0.05; χ2 = 10.79, d.f. = 2, p = 0.005; respectively). Thus, we used a mixed model in which individuals, years and meadows were specified as random factors to test for treatment and infestation effects on juvenile survival. Survival rate was an average for each litter, to avoid pseudo-replication. Juvenile survival was not significantly different for experimental versus control mothers (table 2; χ2 = 0.01, d.f. = 1, p = 0.92). In addition, there was no significant difference in survival rate of young from flea-infested versus -uninfested nests (χ2 = 0.20, d.f. = 1, p = 0.66).
There was no significant difference in survival to the following year between control females and experimental females (χ2 = 0.00, d.f. = 1, p = 0.99). Adult female survival rates did, however, differ significantly between meadows (χ2 = 25.17, d.f. = 3, p < 0.0001). Thus, we used a mixed model in which individuals and meadows were specified as random factors, and a binomial distribution was applied, to test for treatment and infestation effects on adult survival. Adult survival did not differ significantly between treatments (table 2; χ2 = 0.10, d.f. = 1, p = 0.75). Mothers with nest burrows of young that were infested with fleas tended to have better annual survival than mothers with uninfested young (χ2 = 3.29, d.f. = 1, p = 0.07).
(e) Fitness measures
We estimated annual fitness when offspring were yearling age, in the year following the ectoparasite removal treatment. Thus, annual fitness reflects the effect of maternal survival, as well as both production and survival of offspring. None of the mothers for whom we had annual fitness were measured more than once. Annual fitness did not vary significantly between the two years in which we gathered data, but did vary significantly among the three meadows (χ2 = 0.44, d.f. = 1, p = 0.51; χ2 = 23.48, d.f. = 2, p < 0.0001; respectively). Thus, we applied a mixed model in which meadow was applied as a random factor. We found no significant effect of treatment on annual fitness (table 1b; log-likelihood test for treatment effect: χ2 = 0.06, d.f. = 1, p = 0.80). There was no significant difference in annual fitness between mothers with and without nests of flea-infested young (χ2 = 1.29, d.f. = 1, p = 0.26).
(f) Treatment comparison
In a single colony and year, we treated adult females with either Stonghold® solution (N = 6), flea powder (N = 8) or the sham treatment (N = 8). No significant difference among the three treatments was found regarding female reproductive success (Kruskal–Wallis test: χ2 = 1.2, d.f. = 2, p = 0.53), female weight at weaning (Kruskal–Wallis test: χ2 = 0.23, d.f. = 2, p = 0.88) or average offspring weight (Kruskal–Wallis test: χ2 = 1.5, d.f. = 2, p = 0.50).
(g) Power analyses
Given our sample sizes (table 1), an increase in litter size of about 0.47 young or more, or about 24% for experimental females, would have been significant at the 0.05 probability level (t = 1.69, N = 58 experimental females, 44 controls). For annual fitness, an increase of about 0.32, or about 32% for experimental females, would have been significant (t = 1.69, N = 28 experimental mothers, N = 26 controls). Given our sample of litters, an increase in juvenile survival to about 0.381 (compare to control litters in table 2), or about 10% points, would have been significant (binomial one-tailed test, N = 63 experimental litters). For survival of mothers, an increase in survival to about 0.816, or about 14% points, would have been significant (binomial one-tailed test, N = 38 experimental mothers).
We tested the effects of ectoparasite removal on life-history traits of female Columbian ground squirrels over three years and across four colonies. Contrary to our expectations, we found little or no effect of ectoparasite removal either directly on females or for their offspring, regarding: body condition, reproductive success, survival and annual fitness. Overall, our findings suggest that female and juvenile Columbian ground squirrels were largely unaffected by fleas in the years of our experiments.
The predicted positive influence of ectoparasite removal on female Columbian ground squirrels could not be confirmed in this study, which conflicts with the results of a previous experiment on the same species by Neuhaus . That study found that females treated with ectoparasite removal were significantly heavier when they weaned their litters and had significantly larger litters than control mothers. There are several possible explanations for the different findings of how ectoparasites can affect female Columbian ground squirrels. First, in our study a spot-on solution was applied in addition to the flea powder used by Neuhaus . Although this is an agent mainly used for domestic pets (e.g. [42,43]), we cannot exclude a negative effect such as killing of useful intestinal flora. A spot-on solution (based on ‘Selamectin’) has been used on wild common voles (Microtus arvalis) housed in captivity, and no negative effect was reported . In addition, we found no evidence of separate influences of the spot-on solution, flea powder, or sham massage on the reproductive and survival variables of Columbian ground squirrels. Furthermore, we showed that the anti-parasite treatments reduced the ectoparasite load in females (figure 1). A previous study in males demonstrated that in addition to removal of ectoparasites, endoparasites were considerably decreased but not eliminated by applying the spot-on solution . Thus, endoparasites might also have been reduced in experimental females. Taken all together, it is unlikely that the different ectoparasite removal treatments used in the studies caused the inconsistent results between our study and Neuhaus' earlier study .
An alternative explanation might be that the ectoparasites were not present in sufficient numbers for their removal to have resulted in visible improvements in reproduction and survival for the ground squirrels. Even though the ectoparasite removals were successful, the number of observed fleas in the four populations over three years was in general low (figure 1). Females infested with extremely high flea loads were rare among both control and experimental individuals. Thus, females might have easily coped with natural loads of ectoparasites and the infestations were not harsh enough to produce effects on the fitness traits. Columbian ground squirrels are well adapted to deal with very unpredictable environments, such as weather conditions or food availabilities ([45–48], but see ) and might be well adapted to live with naturally occurring levels of parasites. It is possible that the treatment effects (experimental versus control) may have resulted in changes in behaviours, for example, time spent feeding, vigilance or grooming frequencies; but we did not focus on observations of these behaviours. Perhaps when the animals have to deal with other increased costs, such as unfavourable weather conditions, high-density or predator-induced stress levels, an experimental removal or increase in parasite numbers would produce measurable fitness effects. We would expect ectoparasites to be energetically taxing under such conditions, resulting in significantly lower female body weights or declining quality of the offspring; results that were not obtained in this study.
The conflict between the results of the earlier and this study might have arisen because in some years and in some populations, ectoparasite loads reach a threshold after which the animals are no longer able to compensate (for example by increasing their food intake), forcing them to subsume costs into their own condition or those of their offspring. However, in the Neuhaus study , litter sizes were high for both experimental and control females, at 4.33 (±0.18 s.e., n = 9) weaned offspring (females that failed to wean offspring were not counted). Among our four populations, we found an average litter size of 2.57 (±0.11 s.e., n = 79) for females that weaned litters, a litter size closer to the mean of 2.9 for nearby populations (e.g. ). Spring body weight of mothers at emergence from hibernation was also relatively higher in the Neuhaus study , by about 25% (i.e. about 467 g, compared with 378 g in this study), and greater spring mass is associated with larger litters (e.g. [50,51]). Thus, the condition of the population in the earlier  study might have been unusually high. Parasite fitness may sometimes be highest in hosts that are in good body condition (viz., are ‘profitable hosts’), and heavily infested individuals are not always those in the worst condition . A final possibility is that limited sample sizes (viz., four experimental and five control females) and variation in sampling produced an unusual result in the Neuhaus study . In this study, we had sufficient power from our sample sizes to document effects of the flea removal treatments of the magnitudes found in the earlier study.
A meta-analysis showed that parasite-infested individuals had decreased reproductive success among diverse taxa . For example, reproductive success was increased approximately fourfold in female African ground squirrels (Xerus inauris) that experienced a parasite removal treatment . In diverse bird species, hatching success was negatively affected by parasites (e.g. [52,53]). In our study, reproductive success (litter size) was not enhanced for experimental females. However, energetic costs might be defrayed onto offspring. Another study found that parasite removal for mothers resulted in heavier offspring at weaning in red squirrels (Tamiasciurus hudsonicus) but did not affect condition or survival for the mothers . In our study, only the degree of nest infestation was significantly influenced by treatment, with fewer infested nests of treated mothers compared to those from control mothers. However, neither offspring nor adult females appeared to gain fitness or body condition benefits from the experimental parasite removal or by association with uninfested nests. Females with nests of flea-infested young tended to have a better survival compared to mothers with uninfested nests. Flea-infested juveniles, however, showed little difference in survival compared with juveniles from non-infested nests. These results are in line with other studies that did not find an effect of parasites on offspring in body condition or survival. For example, parasite removal from female reindeer (Rangifer tarandus tarandus) did not lead to significantly heavier calves in summer and winter , and direct parasite removal from juvenile Eastern grey kangaroos (Macropus giganteus) had no effect on growth or body mass compared with controls .
An impact of parasite removal on reproduction and breeding success, both often used as fitness proxies, has not always been evident [55,56]. Annual fitness, which measures both production of offspring and the survival of these offspring and their mother to the next active season , was little different between experimental and control females in our study. The annual fitness of mothers with nests and young that were infested with fleas was not significantly different from that of mothers with uninfested young, and the direction of the difference was opposite to that predicted. Thus, fitness estimates also failed to reveal a benefit of ectoparasite removal, or a cost of obvious nest infestations of fleas in Columbian ground squirrels.
The lack of effect of ectoparasites on fitness measures of Columbian ground squirrels might be explained if females were usually less infested than males. Often males are more heavily parasitized than females (e.g. [29,30,57,58]) owing to sexual size dimorphism that favours males [7,28], immunosuppressive effects of androgens that decrease parasite resistance [59,60] or male territorial behaviour associated with larger home ranges [61,62]. However, a previous study of male Columbian ground squirrels showed that ectoparasite load also had no effect on body mass, reproductive behaviour and fitness. A comparison of numbers of fleas among control males and females after spring emergence and at time period 1 (compare fig. 1 in  with our figure 1) revealed a higher flea prevalence in females compared with males (Pearson χ2 = 17.0, d.f. = 1, p < 0.001). This result is in agreement with a female bias in flea load in Richardson's ground squirrels (Urocitellus richardsonii ) . However, our result has to be treated with caution, because we examined only one type of ectoparasite, and different patterns could emerge when including more parasite species (both ecto- and endoparasites; e.g. [63,64]). Furthermore, females emerge later than males, which might have an impact on parasite activity, for example, through increased flea hatching at higher temperatures. Thus, we found a sex-bias in ectoparasites, but further studies are needed to obtain a more complete picture by including more parasite species.
While many studies have shown that parasites have serious negative consequences for hosts (e.g. [22,65–68]), there is increasing evidence from studies that show few or no effects (rodents: [55–56,69], bats: [24,70], birds: [71,72]). In female Columbian ground squirrels, contradictory results for adult females and their offspring were obtained for populations in the same geographical area, though during different times (this study versus ). We suggest that, in a good year, the costs of parasites that co-evolved over long time with their hosts can be compensated. That is, costs of ectoparasites may be obscured, or the host may able to compensate for costs (increased foraging), when environmental conditions are favourable. Under extremely poor conditions, however, when social or environmental stress (food- or predator-related stress, social stress, adverse weather conditions) adds to these costs, female ground squirrels may yet pay these costs via reduced body condition, poor reproduction and/or poor survival for themselves or their offspring.
The research was conducted under animal use protocols from the Biosciences Animal Care Committee, University of Alberta; the Life and Environmental Sciences Animal Resource Center, University of Calgary and the Institutional Animal Care and Use Committee at Auburn University.
Scripts and data are available from the authors on request, and data are available on Dryad (doi:10.5061/dryad.q5270).
S.R. and P.N. came up with the idea and designed the experiment. S.R., P.N. and F.S.D. collected field data. F.S.D. carried out the statistical analyses and drafted the manuscript, with help from S.R. and P.N. All authors gave final approval for publication.
P.N. was funded by the Swiss National Science Foundation (SNF 3100AO-109816). F.S.D. was funded by the National Science Foundation (DEB-0089473).
We owe special thanks to Peter Kappeler and his colleagues at the Abteilung Verhaltensökologie & Soziobiologie, Deutsches Primatenzentrum, and Abteilung für Soziobiologie/Anthropologie, Johann-Friedrich-Blumenbach Institut für Zoologie & Anthropologie, Universität Göttingen, for inviting us to present our study at the Göttinger Freilandtage conference ‘The Sociality–Health–Fitness Link’ in 2013. P. M. Kappeler, J. O. Murie, C. L. Nunn and two anonymous referees kindly made suggestions for improvements to the manuscript. For help in the field, we thank A. Balmer, M. Berger, C. Deleglise, E. Emery, B. M. Fairbanks, C. Grossen, C. Heiniger, L. Hofmann, S. Röösli and A. L. Skiebiel. Housing at the R. B. Miller Field Station during the field season was provided by the University of Calgary's Biogeosciences Institute; we thank Station Manager J. Buchanan-Mappin, Institute Director E. Johnson, and Field Station Responsible K. Ruckstuhl for their support. Furthermore, we thank Pfizer Animal Health Canada for generously providing the project with their product Stonghold®.
One contribution of 14 to a theme issue ‘The sociality–health–fitness nexus in animal societies’.
- Accepted December 2, 2014.
- © 2015 The Author(s) Published by the Royal Society. All rights reserved.