INTRODUCTION
Variability among host individuals in the number of parasite species they harbour, and in the severity of infection by these species, is a universal feature of any host population (Poulin, Reference Poulin2007). This variability is the outcome of several processes: heterogeneous distribution of infective stages and their chance encounter with hosts, differences among individual hosts in their exposure or susceptibility to infection, as well as direct and indirect interactions among parasites within the host that may have positive or negative effects on infections by certain species (Poulin, Reference Poulin2001, Reference Poulin2007).
Studies on rodent hosts have demonstrated the action of these processes. For instance, interactive associations between pairs of helminth species among host individuals have been observed repeatedly in both laboratory (Holland, Reference Holland1984) and field studies (Haukisalmi and Henttonen, Reference Haukisalmi and Henttonen1993; Behnke et al. Reference Behnke, Gilbert, Abu-Madi and Lewis2005). Interactions between parasite species are not always negative; there is evidence that infection by one species can facilitate infection by others via a decrease in host immune responses (Behnke et al. Reference Behnke, Eira, Rogan, Gilbert, Torres, Miquel and Lewis2009). These processes, combined with chance exposure to patchily distributed infective stages, would be enough to generate much variation in parasite species richness and infection levels among individual hosts. In addition to these factors, however, there exist intrinsic differences in past exposure and susceptibility to parasites among individual hosts; for instance, older hosts have been exposed for longer and should harbour more parasites than young ones, and there may also be behavioural or physiological differences between host sexes influencing infection patterns (Behnke et al. Reference Behnke, Lewis, Zain and Gilbert1999; Eira et al. Reference Eira, Torres, Vingada and Miquel2006). In particular, host sex is considered to be important, because the immunosuppression induced by testosterone generally causes males to harbour more parasites than females (Poulin, Reference Poulin1996; Schalk and Forbes, Reference Schalk and Forbes1997). In rodents, this often leads to males, especially older males, harbouring a substantial portion of the parasite population, and releasing disproportionately more infective stages into the environment than females (Ferrari et al. Reference Ferrari, Cattadori, Nespereira, Rizzoli and Hudson2004; Luong et al. Reference Luong, Grear and Hudson2009). Females may also experience immunosuppression during breeding or lactation, at which times their infections may be higher than those of non-reproducing females (Vandegrift and Hudson, Reference Vandegrift and Hudson2009).
In some host species, however, unusual ecological conditions may either exacerbate differences in parasite richness and infection levels among individual hosts, or homogenize infections such that inter-individual differences become very small. In subterranean herbivorous rodents of the South American genus Ctenomys (Octodontidae), for instance, individuals live singly in permanently sealed burrow systems and in low-density populations (Malizia et al. Reference Malizia, Vassallo and Busch1991; Malizia, Reference Malizia1998). They only emerge occasionally from their burrows for brief surface excursions to collect plant material, although they also feed on below-ground parts of plants and may engage in coprophagy (Busch et al. Reference Busch, Malizia, Scaglia and Reig1989; del Valle et al. Reference del Valle, Lohfelt, Comparatore, Cid and Busch2001). Pregnant females and those with pups make even fewer surface visits than other members of the population, and young females remain with their mothers almost until maturity (Malizia, Reference Malizia1998). Individuals place fecal pellets outside the plugged entrance of their burrow to signal that it is occupied, and they will inspect fecal deposits from neighbouring burrows during their occasional surface excursions (Fanjul et al. Reference Fanjul, Zenuto and Busch2003); this may provide opportunities for parasite transmission. The extremely sedentary nature of these fossorial rodents, with most of the time spent in the underground burrow and little contact with conspecifics, may affect their exposure to parasites and impact on inter-individual variation in infections.
There have only been a few previous parasitological studies of Ctenomys rodents (Rossin and Malizia, Reference Rossin and Malizia2002, Reference Rossin and Malizia2005; Rossin et al. Reference Rossin, Timi and Malizia2004, Reference Rossin, Poulin, Timi and Malizia2005 a, Reference Rossin, Timi and Maliziab, Reference Rossin, Timi and Malizia2006 a, Reference Rossin, Timi and Maliziab, Reference Rossin, Varela and Timi2009) or other fossorial rodents (e.g., Scharff et al. Reference Scharff, Burda, Tenora, Kawalika and Barus1997). Here, we focus on the determinants of infection by separate helminth species in 2 species, C. australis and C. talarum. Specifically, we examined (i) whether a male bias is observed in infection levels in these 2 species characterized by sexual size dimorphism, with males being 30–50% heavier than females (Malizia et al. Reference Malizia, Vassallo and Busch1991), and (ii) what kinds of inter-specific associations exist among helminth species.
MATERIALS AND METHODS
Hosts were caught across seasons at 2 localities. Specimens of C. talarum were caught at Mar de Cobo, Buenos Aires Province (37°58′S, 57°34′W) during 2000–2001, whereas those of C. australis were caught at Necochea, Buenos Aires Province (38°33′S, 58°45′W) during 2003–2004. The distribution of both species is linked to micropatterns of soil and vegetation; C. australis inhabits areas with sparse vegetation and deep sandy soils, while C. talarum lives in areas with dense vegetation, and compact and shallow soils (Malizia et al. Reference Malizia, Vassallo and Busch1991; Comparatore et al. Reference Comparatore, Agnusdei and Busch1992). In the field, inhabited burrow systems were distinguished by the presence of conspicuous mounds of fresh soil brought to the surface during burrowing activities of the rodents. Live traps were placed in these burrows close to their entrance, and checked every hour throughout the day during each trapping session. Trapped rodents were killed by over-exposure to ether and returned to the laboratory for measurements and dissection for parasites.
For each host individual, the following information was recorded: (i) date of capture, with dates subsequently grouped into seasons; (ii) body length (cm), excluding the tail; (iii) body mass (g), including foetuses in the case of pregnant females; (iv) age in years, determined based on the extent of epiphyseal ossification of the humeri (Malizia and Busch, Reference Malizia and Busch1991); (v) reproductive condition, i.e. immature or mature; (vi) sex, with females separated into pregnant and non-pregnant; and (vii) the identity, number and location of each endohelminth.
Since variables like age, body length, body mass, sex and reproductive condition are correlated with each other and thus not statistically independent, we collapsed them into 3 predictor variables that were independent and still captured all the relevant biological information. They were: (i) age in years, (ii) sex, which included the 3 categories--males, non-pregnant females and pregnant females, and (iii) body condition, which corresponded to the residual value of body mass regressed against body length. These regressions were significant (Ctenomys australis: mass=26·6*length−420·7, r 2=0·64, F 1,43=79·1, P<0·0001; C. talarum: mass=7·3*length−35·4, r 2=0·47, F 1,79=72·1, P<0·0001), and positive residuals indicate animals heavier than average for their length, whereas negative values indicate animals with a lower mass than expected for their body length.
We used generalized linear models performed in the R environment (version 2.9.1; R Development Core Team, 2009) to evaluate the independent effect of multiple predictor variables on infection by each parasite species, separately for each host species. For common (prevalence ⩾48%) parasite species, the response variable was abundance of infection, i.e. the number of individual parasites per host (with uninfected hosts included in the analysis), and we used a quasipoisson error structure with a log-link function. A quasipoisson error structure provided a better fit (based on deviance) between models and data than the negative binomial, although using the former or the latter generally yields the same results in a GLM (see Crawley, Reference Crawley2007, pp. 556–558). For relatively rare parasites (prevalence <30%), the response was a binary variable, i.e. presence or absence of the parasite, and we used a binomial error with complementary log-log link function (Crawley, Reference Crawley2007). In all GLMs, the predictor variables were season of capture (4 seasons), age (continuous variable, 1–6 years), sex (males, non-pregnant females, pregnant females), body condition, and infection by other parasite species, treated either as a continuous variable (abundance) or as a binary one (presence or absence) for rare species. Exploratory analyses, if necessary with reduced models (i.e. fewer factors in order to avoid empty cells), indicated that second-order interactions were all non-significant, and therefore they were not included in the final models. Higher-order interactions (three-way and above) could not be evaluated because of limited sample sizes.
RESULTS
Overall, 45 Ctenomys australis (2–6 years of age, 78–530 g in body mass, 13 non-pregnant females, 13 pregnant females, and 19 males) and 81 C. talarum (1–6 years of age, 48–187 g in body mass, 25 non-pregnant females, 17 pregnant females, and 39 males) were included in the study. Except for cysts of the cestode Taenia talicei found in the body cavity of both rodent species, all other helminths recovered were nematodes associated with the digestive tract (Table 1). All parasites are acquired by ingestion, except Strongyloides myopotami which has skin-penetrating infective stages.
Table 1. Summary of the parasite species and their infection parameters in the two rodent host species
In C. australis, both the cestode T. talicei and the nematode Pudica ctenomydis were not common (Table 1); they were included as predictor variables in the GLM summarized below, but not as response variables, i.e. their patterns of infection were not the subjects of separate GLMs. Only Trichuris pampeana was subject to an analysis, and of all predictors considered, only sex (sex [males]: change in coefficient estimate=−0·713, t=−2·07, P=0·0458) had a significant effect on its abundance of infection in the GLM. Males tended to harbour fewer nematodes of this species than pregnant and non-pregnant females (Fig. 1).
Fig. 1. Mean (±s.e.) number of the nematode Trichuris pampeana per host in the rodent Ctenomys australis as a function of the sex and reproductive status of the hosts. Numbers above bars are the number of hosts per category.
In C. talarum, the nematode Trichostrongylus duretteae occurred at very low prevalence and abundance of infection (Table 1), and it was therefore excluded from all analyses. All other species were included as predictors of the presence or abundance of other species, and all were the response variable in their own dedicated GLM. Some significant effects emerged from those analyses (Table 2). There were seasonal effects in 2 parasite species, both of which were more likely to infect hosts during the warmer spring and summer months. Two parasite species were positively correlated with host age, although host body condition had no effect on the presence or abundance of any of the parasite species (Table 2). One species, Strongyloides myopotami, was less likely to occur in males than in females (Table 2; Fig. 2). Finally, there were some significant associations among pairs of parasite species. First, there was a negative association between Pudica ctenomydis and Graphidiodes subterraneus, although only significantly affecting the latter. Second, there was a positive association between G. subterraneus and S. myopotami, although again it was not symmetrical and only significant for the latter species (Table 2). Third, there was a significant positive association between the presence of P. ctenomydis and that of S. myopotami. Finally, there was a strong positive association between the presence of G. subterraneus and the abundance of Taenia talicei across hosts (Fig. 3); hosts infected by G. subterraneus harboured over 3 times more T. talicei, on average, than those not infected by G. subterraneus.
Fig. 2. Prevalence (±95% confidence interval), i.e. percentage of individuals infected, of the nematode Strongyloides myopotami per host in the rodent Ctenomys talarum as a function of the sex and reproductive status of the hosts. Numbers above bars are the number of hosts per category.
Fig. 3. Mean (±s.e.) number of the cestode Taenia talicei per host in the rodent Ctenomys talarum as a function of whether or not the host harboured the nematode Graphidiodes subterraneus. Numbers above bars are the number of hosts per category.
Table 2. Results of GLMs evaluating the effects of several factors on either the abundance or presence of six parasite species in the rodent host Ctenomys talarum
(Data shown are changes in coefficient estimates (compared to intercept; significant ones in bold) when that factor alone is excluded from the model; significance is based on a t-test.)
* P<0·05; ** P<0·01; *** P<0·001.
DISCUSSION
The unusual solitary and subterranean lifestyle of Ctenomys rodents may create patterns of infections among individual hosts that could depart from those observed in other rodents living above ground and with greater likelihood of interactions with conspecifics. We specifically looked at possible sex-biases in infection patterns, and at statistical associations among parasite species, to determine to what extent host characteristics and/or interspecific interactions among parasites might structure the helminth communities in Ctenomys australis and C. talarum.
Of the 7 tests of sex bias we performed (1 helminth species in C. australis and 6 in C. talarum), 2 showed a significant pattern. Unlike the widely reported male-bias that is particularly common in nematode infections of mammals (Poulin, Reference Poulin1996; Schalk and Forbes, Reference Schalk and Forbes1997) and generally attributed to testosterone-induced immunosuppression (Zuk and McKean, Reference Zuk and McKean1996), we observed a clear female bias in both cases. This arose despite the sexual size dimorphism in these species, where males are significantly larger than females (Malizia et al. Reference Malizia, Vassallo and Busch1991). In the case of the nematode Strongyloides myopotami in C. talarum, the bias could be attributed in part to the behaviour of females, which spend more time in their burrows than males, both while pregnant and after giving birth. Since this nematode has free-living infective stages that penetrate host skin, the virtual immobility of females for prolonged periods could facilitate re-infection as well as mother-to-daughter transmission, since the latter have high phylopatric tendencies, staying in their mother's burrow for long periods (Malizia et al. Reference Malizia, Zenuto and Busch1995). These behaviours specific to females should enhance the probability of infection. The same behaviours may also predispose C. australis females to higher infections by the nematode Trichuris pampeana, although if this is the case we might expect the other parasites acquired by ingestion to show female-biased infections, and they do not. Breeding, from pregnancy to the end of lactation, may also cause immunosuppression in females (Zuk and McKean, Reference Zuk and McKean1996), which can lead to increased infections (e.g. Vandegrift and Hudson, Reference Vandegrift and Hudson2009). Our results for both cases show that non-pregnant females had higher infections than pregnant ones. Because of our method of capture, it was not possible to determine whether or not non-pregnant females were lactating pups at the time of sampling. Although some non-pregnant females had well-developed mammary glands surrounded by areas where the fur has been lost, this is not a sure sign of lactation, and the entire burrow could not be excavated to search for pups. Nevertheless, the 2 sex biases we observed join a short list of exceptions (Dick et al. Reference Dick, Gannon, Little and Patrick2003; Zahn and Rupp, Reference Zahn and Rupp2004; Krasnov et al. Reference Krasnov, Morand, Hawlena, Khokhlova and Shenbrot2005) that go against the general pattern in which males are more frequently or heavily infected than females; the reversed pattern observed in subterranean Ctenomys rodents may be a consequence of their unusual lifestyle.
Our generalized linear models also revealed other effects on infections. Indeed, host age and season of capture both influenced infection levels of some helminth species, as seen in other rodent-parasite systems (Behnke et al. Reference Behnke, Lewis, Zain and Gilbert1999; Eira et al. Reference Eira, Torres, Vingada and Miquel2006; Vandegrift and Hudson, Reference Vandegrift and Hudson2009). These effects are neither unusual nor surprising. Older hosts have had longer to acquire and accumulate parasites, and we found a positive effect of host age on infection in 2 species. Warmer parts of the year are often associated with higher rates of infection because hosts are more active and feed more than during cold periods. Summer peaks in nematode infections have been reported in geomyid pocket gophers from North America, another taxon of subterranean rodents (Gardner, Reference Gardner1991). In the case of the nematode Strongyloides myopotami in C. talarum, it is also likely that the skin-penetrating infective larvae are more active during spring and summer months, and thus more likely to contact host bodies. The one factor that did not affect infections in our study systems was host body condition, measured here as observed body weight relative to that expected for a given body length. This suggests that hosts that fed more did not automatically acquire more ingestion-transmitted parasites, and that heavily-infected hosts did not incur reductions in body condition as a consequence of infection.
We also uncovered statistically significant associations between parasite species among the helminths infecting C. talarum. Our generalized linear models, because they take into account other species as well as other confounding factors, should yield more robust tests of pairwise associations between parasite species than many earlier correlation tests (see Haukisalmi and Henttonen, Reference Haukisalmi and Henttonen1998; Poulin, Reference Poulin2005). Of the 4 pairwise associations we found 2 were asymmetric, i.e. one species affected the presence or abundance of another, but not vice versa. This is actually a common phenomenon in helminth communities (see Poulin, Reference Poulin2007, for review). Also, pairs of significantly associated species consisted, in 3 of the 4 cases, of species living in different microhabitats within the host, suggesting that immune-mediated effects are more likely explanations than active resource competition. Recent studies have emphasized that positive interactions are as frequent, if not more frequent, than negative ones in helminth communities of rodents and other mammals (Behnke et al. Reference Behnke, Gilbert, Abu-Madi and Lewis2005, Reference Behnke, Eira, Rogan, Gilbert, Torres, Miquel and Lewis2009; Lello et al. Reference Lello, Boag, Fenton, Stevenson and Hudson2004). For instance, Behnke et al. (Reference Behnke, Eira, Rogan, Gilbert, Torres, Miquel and Lewis2009) have provided data suggesting that the nematode Heligmosomoides polygyrus facilitates infection of its rodent host, Apodemus sylvaticus, by other helminths by suppressing parts of its immune response. In our study, 3 of 4 significant associations were positive, suggesting either some form of facilitation as in the preceding example, or that both species are acquired together in a way that generates a statistical association. However, in the case of the positive association between the nematodes Pudica ctenomydis and Strongyloides myopotami, some form of active facilitation seems likely, since the two species have different modes of infection: the former species in the pair is acquired by ingestion, whereas the latter penetrates host skin. Interestingly, a relative of S. myopotami, the well-studied S. ratti, is very sensitive to host immune status, and any helminth that could suppress host responses should concomitantly facilitate S. ratti infections (Wilkes et al. Reference Wilkes, Thompson, Gardner, Paterson and Viney2004).
Overall, a range of factors combines to determine the presence or abundance of infection of the various helminths found in the 2 Ctenomys species. These range from the season of capture to interactions among parasite species. In particular, we found that when differences in infections exist between host sexes, they go against the general trend for nematodes in mammals, and show a female bias that may be a direct consequence of the solitary and subterranean life of these rodents compared to most other host species studied to date. These results argue for a strong role of ecology in determining infection patterns, even to the extent of overshadowing physiological differences between hosts.
ACKNOWLEDGEMENTS
We thank Dr M. C. Del Valle (Laboratorio de Ecofisiología, Universidad Nacional de Mar del Plata, Argentina) for her help during trapping sessions in the field.
