Infections do not predict shedding in co-infections with two helminths from a natural system

Transcription

1 Ecology, 95(6), 2014, pp Ó 2014 by the Ecological Society of America Infections do not predict shedding in co-infections with two helminths from a natural system ISABELLA M. CATTADORI, 1,3 BENJAMIN R. WAGNER, 1 LAURA A. WODZINSKI, 1 ASHUTOSH K. PATHAK, 1 ADAM POOLE, 1 AND BRIAN BOAG 2 1 Center for Infectious Disease Dynamics and Department of Biology, Pennsylvania State University, University Park, Pennsylvania USA 2 The James Hutton Institute, Invergowrie, Dundee DD2 5DA United Kingdom Abstract. Given the health and economic burden associated with the widespread occurrence of co-infections in humans and agricultural animals, understanding how coinfections contribute to host heterogeneity to infection and transmission is critical if we are to assess risk of infection based on host characteristics. Here, we examine whether host heterogeneity to infection leads to similar heterogeneity in transmission in a population of rabbits single and co-infected with two helminths and monitored monthly for eight years. Compared to single infections, co-infected rabbits carried higher Trichostrongylus retortaeformis intensities, shorter worms with fewer eggs in utero, and shed similar numbers of parasite eggs. In contrast, the same co-infected rabbits harbored fewer Graphidium strigosum with longer bodies and more eggs in utero, and shed more eggs of this helminth. A positive density-dependent relationship between fecundity and intensity was found for T. retortaeformis but not G. strigosum in co-infected rabbits. Juvenile rabbits contributed to most of the infection and shedding of T. retortaeformis, while adult hosts were more important for G. strigosum dynamics of infection and transmission, and this pattern was consistent in single and co-infected individuals. This host parasite system suggests that we cannot predict the pattern of parasite shedding during co-infections based on intensity of infection alone. We suggest that a mismatching between susceptibility and infectiousness should be expected in helminth coinfections and should not be overlooked. Key words: eggs in host s feces; gastrointestinal helminths; Graphidium strigosum; host age; intensity of infection; Oryctolagus cuniculus; parasite body length; parasite eggs in utero; rabbit; single and dual infections; Trichostrongylus retortaeformis. INTRODUCTION Parasites play a critical part in food web dynamics and biodiversity and represent a major health and economic problem for humans and livestock (Over et al. 1992, Geerts and Gryseels 2000, Hudson et al. 2006, Hotez 2008, Lafferty et al. 2008, Lustigman et al. 2012). An integrated understanding of the patterns of infection and shedding at the host level is necessary if we are to appreciate ecological processes influencing disease transmission as well as for improving individual-based intervention strategies. Theory predicts that co-infections can increase the likelihood of parasite invasion and spread in a host population by promoting positive co-variations between host susceptibility and infectiousness (i.e., the ability of an infected individual to infect naive cases) (Graham et al. 2007). By doing so, co-infections can augment individual variation to infection where hosts with multiple infections carry and transmit the majority of Manuscript received 8 August 2013; revised 4 November 2013; accepted 15 November Corresponding Editor: D. M. Tompkins. 3 imc3@psu.edu 1684 infections to the point that they become critically important in seeding the next outbreak, for instance, by developing into super-shedding and/or super-spreading cases (Lloyd-Smith et al. 2005, Chase-Topping et al. 2008, Garske and Rhodes 2008, Lass et al. 2013). The opposite scenario can also be true where, by removing the heavily co-infected individuals, co-infections can reduce host heterogeneity to infection, stabilize parasite host dynamics and thus decrease the risk of unpredictable outbreaks (Fenton 2008). The general consensus among infectious disease studies is that heavily parasitized, co-infected hosts also transmit more infective stages (e.g., helminth eggs, fungal spores, bacterial or viral particles) than individuals with single infections. However, whether this high burden will result in high transmission or supershedding/spreading events still remains controversial (Kao et al. 2007, McCullers et al. 2010, Lass et al. 2013). Part of the issue is that host parasite interactions often exhibit nonlinear relationships, and transmission is strongly affected both by host characteristics (i.e., age, immuno-physiological status, and behavior, among some) and parasite properties such as survival, development, and fecundity. Furthermore, factors that influence

2 June 2014 SHEDDING IN PARASITE INFECTIONS 1685 parasite establishment and survival might have a different impact on fecundity or shedding, and the level of these relationships can vary both between and within hosts during co-infections. Therefore, assuming that high parasitic infections are indicative of high shedding loads and transmission needs to be confirmed in coinfection settings. For parasitic helminths, this can be determined by assessing how the presence of a second infection (either micro- or a macro-parasite) alter the relationship between worm intensity, development and fecundity, and how these changes modulate the number of eggs shed into the environment. Importantly, by identifying which co-infected individuals are responsible for the majority of eggs shed, and how they differ from singleinfected cases, we can provide fundamental details on risk of infection and transmission based on hosts in different infection settings. The simple prediction is that shedding proportionally increases with the infection such that the higher the parasite intensity, the more infective stages are shed into the environment, where female per capita fecundity/shedding remains unchanged (Lass et al. 2013). However, co-infections can also influence vital rates and life history strategies, where parasites may accelerate sexual maturity (Lagrue and Poulin 2008), grow bigger, have higher fecundity, and shed more per capita infective stages than in singleinfected hosts. There again, negative associations between infection and shedding, caused by host and parasite constraints, might alter parasite dynamics such that a high infection results in reduced per capita fecundity. Therefore, the total number of eggs or larvae shed at any time into the environment by an infrapopulation of parasites may remain fundamentally unchanged or show small variation between single and co-infected hosts. Nonlinear relationships can create additional levels of complexity. For example, coinfections might boost shedding below a parasite threshold but reduce it above this critical density. Studies on single helminth infections indicate that parasite development and fecundity are frequently inversely related to parasite abundance (Keymer and Slater 1987, Quinnell et al. 1990, Anderson and May 1992), supporting the negative intensity-shedding hypothesis. While we should expect similar trends in coinfections, changes in host susceptibility may modify these relationships to the point that we cannot predict the infection transmission relationship based on findings from single infections alone (Holmes 1961, 1962, Lagrue and Poulin 2008, Pathak et al. 2012, Thakar et al. 2012). In this study, we examine whether host heterogeneity to infection leads to a proportional variation in host transmission by investigating how concurrent infections with two gastrointestinal helminths influence shedding (i.e., eggs in host s feces), eggs in utero, and body length of both species in an herbivore host. Natural infections of the European rabbit (Oryctolagus cuniculus) with Trichostrongylus retortaeformis increase with host age, peak in juveniles, and decrease in old rabbits, while Graphidium strigosum (see Plate 1) accumulates with host age (Cattadori et al. 2005, 2008). We previously found that in rabbits co-infected with the two worms, T. retortaeformis intensity and aggregation were higher than in single-infected individuals, while G. strigosum intensity was unchanged, although aggregation decreased (Cattadori et al. 2008). Laboratory and modeling studies showed that host immunity mainly regulates the dynamics of T. retortaeformis infection (Cornell et al. 2008, Murphy et al. 2011, 2013, Thakar et al. 2012), but there is a weak immune control of G. strigosum (Murphy et al. 2011, 2013). Based on these findings and using a new data set from an independent rabbit population, we predict that co-infection with the second helminth will modify T. retortaeformis intensity and number of eggs in utero, while no apparent effects will be observed for G. strigosum characteristics. We also predict that, despite these demographic changes, the amount of parasite eggs shed, at any point in time, will not differ between single and co-infected individuals for either helminth species. MATERIAL AND METHODS Host and parasite data European rabbits (Oryctolagus cuniculus) were collected monthly from 2004 to 2011 in an agro-ecosystem in Perthshire, Scotland (UK) according to UK regulations and field procedures approved by the Institutional Animal Care and Use Committee of The Pennsylvania State University. A total of 1865 rabbits were sampled (interannual sample range: rabbits, P. 0.05), 834 females and 1031 males (P, 0.05), individuals across age classes, from 2 to 8þ month-old cases: 228, 185, 197, 300, 513, 392, and 47 rabbits, respectively (P. 0.05). Host age was recorded following procedures we have described elsewhere (Cattadori et al. 2005, 2008). Briefly, rabbits were grouped in seven classes of onemonth step increment, from two-month-old kittens to 8þ month-old adults. Intensities of T. retortaeformis infection in the small intestine and G. strigosum in the stomach were quantified by inspecting 10-mL aliquot counts from 4% to 50% of the organ content (Boag 1972). The parasite female-to-male ratio was estimated in 2007 and 2008, and the sex ratio (T. retortaeformis ¼ and G. strigosum ¼ ; mean 6 SE,) was used to calculate the total number of females of both species in every rabbit. For each individual, a random subsample of parasites (5 10 specimens for each sex) was collected and stored in 0.4% formalin. Parasite body length was measured using a camera (Infinity-1; Lumenera Corporation, Ottawa, Ontario, Canada), connected to a stereo-microscope and a PC (software: Image-J, NIH) (Chylinski et al. 2009). The total number of eggs in a T. retortaeformis female s uterus was directly counted under the stereo-microscope, while G. strigosum

3 1686 ISABELLA M. CATTADORI ET AL. Ecology, Vol. 95, No. 6 females were digested, eggs extracted, and subsequently counted under the stereo-microscope (Hussey and Barker 1973). Eggs counted in utero were used as indicative of per capita fecundity at time of sampling. To quantify the parasites shed into the environment, 2 g of pellets (feces) were collected from the rabbit s colon, and the number of eggs per gram of feces (Gordon and Whitlock 1939) was estimated for every host sampled during the period. The two helminths have very similar eggs, both in size (G. strigosum, ;100 lm and T. retortaeformis, ;90 lm) and shape (Taylor et al. 2007), and are difficult to distinguish under the microscope. Therefore, the number of the two worms shed by co-infected rabbits was estimated using parasite fecundity and intensity of female infection. Initially, the ratio of eggs shed by each parasite from a co-infected rabbit was estimated as GSe it 3 GSf it TRe it 3 TRf it ; ð1þ where GS is G. strigosum and TR is T. retortaeformis, e is the mean number of eggs in utero and f the total number of infecting females, for every host i at a given time t (i.e., day of sampling). This represents the potential number of eggs shed, since they have not left the female s body. The number of eggs shed in 1 g of fecal material (epg), by each parasite species in a coinfected rabbit, is then and GSepg it ¼ GSe it 3 GSf it Totalepg 3 it TRe it 3 TRf it 1 þ GSe it 3 GSf it TRe it 3 TRf it TRepg it ¼ Totalepg it GSepg it : ð2þ It is important to note that eggs in utero and shed, by single and co-infected rabbits, represent a snapshot on a given sampling day and do not provide information on the per capita parasite fecundity rate or rabbit shedding rate over the course of the infection. Changes in food intake, as well as quality and rate of feeding, influence defecation frequency and volume (Thacker and Brandt 1955), making it difficult to estimate the longitudinal rate of shedding in the field. Our estimation can be considered as indicative of the average amount of eggs shed by a rabbit of identified age, intensity of infection, and at any day during the specified month of sampling. Finally, blood was collected from every individual sampled between 2005 and 2010 and serum speciesspecific antibody response quantified to have an indication of changes in the immune response between single and co-infected hosts. Antibody detection To estimate the host immune response to the two parasites, systemic antibodies IgA and IgG were detected using adult worm excretory/secretory (ES) products as a source of antigen and enzyme-linked immunosorbent assays (ELISAs) adapted from our previous work (Murphy et al. 2011, 2013). This approach provided a robust measure of species-specific antibody responses against the two helminths in wild rabbits. Methodological details for the ES preparation and antibody detection are reported in the Appendix. Antibody IgA and IgG have been suggested to play a key role in helminth development and fecundity (Stear et al. 1995, Henderson and Stear 2006, Paterson and Barber 2007). We recently showed that IgA and IgG contributed to T. retortaeformis reduction from the small intestine, indicating that they are fundamental but not sufficient for controlling this parasite (Thakar et al. 2012). IgA and IgG also contribute to the immune response to G. strigosum, although values are generally low in the stomach mucus (Murphy et al. 2011, 2013). Statistical analysis To characterize the relationships between parasite attributes (i.e., intensity of infection, body length, eggs in utero, and eggs in feces) and whether these relationships differed between single and co-infected individuals, generalized linear models (GLM with Poisson error distribution or negative binomial error distribution) or generalized linear mixed models (GLMM with Poisson error distribution Lme4 package [Pinheiro and Bates 2000]) were implemented. GLM was also used to identify the host age class responsible for the majority of shedding and infection, and whether these age classes matched between single and co-infected rabbits. Model selection was based on the approach that minimized the AIC value (Burnham and Anderson 1998). For GLMM models, to take into account the nonindependent effect of measuring helminths from the same host, the rabbit s ID code was included as a random effect. Also, to control for a possible seasonal effect and changes in rabbit properties with age (not necessarily caused by age), month of sampling and host age were included as random effects. Rabbit s ID was nested into sampling month and host age, where month of sampling represents the highest level of nesting. Analyses were independently assessed for each parasite, such that G. strigosum was treated as a co-infection when the T. retortaeformis rabbit relationships were examined, and vice versa for G. strigosum. RESULTS T. retortaeformis infections Rabbits single infected with T. retortaeformis harbored fewer parasites, exhibited lower species-specific antibody responses, and carried worms with longer bodies and females with more eggs in utero compared to individuals co-infected with G. strigosum (Fig. 1a, Table 1). Eggs in utero, corrected by worm length and averaged across the rabbits sampled, weakly decreased with

4 June 2014 SHEDDING IN PARASITE INFECTIONS 1687 FIG. 1. The x-axis variables are ln(body length) with the original lengths in millimeters, ln(eggs in utero þ 1), ln(infection intensity), and systemic antibody IgA and IgG response [ln(o.d. þ 1)] of (a) T. retortaeformis and (b) G. strigosum in single (black) and co-infected (gray) rabbits. Means and standard errors are reported. parasite intensity in single infections but increased in coinfected individuals (Fig. 2a, two-way interaction intensity infection type, n ¼ 1368, P, 0.05). The positive trend in co-infected hosts was not caused by the increase of female body length with intensity (Fig. 2c, intensity infection type, P ¼ 0.08), but by the increase of the number of eggs in utero (Fig. 2b, intensity infection type, n ¼ 1370, P, 0.01). Further examination showed that mean number of eggs in utero increased with worm length at a rate that was relatively faster in co-infected than single-infected hosts (Appendix: Fig A1a, length infection type, n ¼ 1369, P, ). These trends were consistent when the analyses were repeated using the whole infra-population of T. retortaeformis females, rather than rabbit averages (Appendix: Figs. A1b, A2a, b, c). By knowing the number of eggs in utero and the proportion of infecting females, we estimated the potential number of T. retortaeformis eggs that could be shed in the environment by every host (both single and co-infected) at the time of sampling. To stress our previous remark, this quantity can be considered representative of the standing crop of eggs potentially shed by a rabbit of a specified age and level of infection at any point during a determined sampling month. Overall, potential shedding was not significantly different between types of infection, but increased with parasite intensity at a rate that was faster in single than co-infected hosts (Appendix: Fig. A3a, intensity infection type, n ¼ 1370, P, ). To examine how well the extent of potential shedding explained the amount of real shedding (i.e., eggs in feces), these two variables were examined in singleinfected rabbits, since eggs in feces are not independent from eggs in utero in co-infected cases as described in Eq. 1. Real and potential shedding were positively associated (Fig. 3a, n ¼ 236, P, , log-linear regression coefficient, 1.05), and while more eggs appeared to be potentially shed than counted in the feces (i.e., all the cases distributed on the right side of the dashed line in Fig. 3a), this was caused by estimating the number of eggs in a representative but small portion of the feces (the standard 1 g of pellets, epg). Interestingly, a fraction of hosts were infected with worms carrying eggs but no eggs were found in the feces at the time of sampling. We then investigated whether intensity of infection could be a robust predictor of shedding in co-infected hosts and whether this differed from single infections. The number of eggs shed in feces was comparable between the two infection types. However, shedding increased with parasite intensity at a rate that was proportionally faster in single than co-infected rabbits (Fig. 4a, intensity infection type, n ¼ 766, P, ). Overall, these findings support the hypothesis that higher T. retortaeformis intensities in co-infected rabbits do not lead to higher shedding loads. To identify the group of rabbits responsible for the majority of parasites shed and whether they also represented the most heavily infected cases, the patterns of shedding and infection were examined among host age classes. Real shedding peaked in juveniles (four months old in co-infected and three months old in single-infected hosts) and decreased in adults (Fig. 5b, n ¼ 766, P, 0.001), but no significant differences were found when type of infection was also considered (i.e., two-way interaction age type of infection, P. 0.05). Generalized linear mixed model (for parasite body length and eggs in utero) and linear mixed models (for parasite intensity and host antibodies) between parasite or host characteristics, as a response, and type of infection as fixed term. TABLE 1. Characteristic T. retortaeformis G. strigosum Coeff. SE N Pr(.jzj) Coeff. SE N Pr(.jzj) Parasite intensity , , Body length , , Eggs in uterus , , Antibody IgA , , Antibody IgG , , Notes: Random effects considered: sampling month, rabbit age, and rabbit ID code. The coefficients represent the co-infected vs. single-infected comparison; a positive coefficient indicates higher values for co-infected than single-infected cases, while a negative value indicates the opposite.

5 1688 ISABELLA M. CATTADORI ET AL. Ecology, Vol. 95, No. 6 FIG. 2. Relationships between parasite characteristics in single (black circles) and co-infected (gray circles) rabbits for (a, b, c) T. retortaeformis and (d, e, f) G. strigosum. Circles represent mean values per rabbit; the lines depict fitted predictions from linear models. The y-axis is ln(body length) with the original lengths in millimeters, ln(eggs in utero þ 1), and the x-axis is ln(intensity of infection) by either parasite. Intensity of infection also peaked in juveniles and decreased in older individuals (quadratic and linear age term for both, n ¼ 1533, P, ) and was consistently lower in single than co-infected hosts across the age classes (Fig. 5a, quadratic and linear age termtype of infection for both, n ¼ 1533, P, ). G. strigosum infections Higher G. strigosum intensities, stronger speciesspecific antibody responses, but shorter parasites with fewer eggs in utero were found in single-infected than in individuals co-infected with T. retortaeformis (Fig. 1b, Table 1). The number of eggs in utero, corrected by female length and averaged among rabbits, decreased with parasite intensity (n ¼ 938, P, 0.001), but no differences were observed between the two types of infection (Fig. 2d). When only the mean number of eggs in utero was considered, these negative trends were still apparent (Fig. 2e, intensity type of infection, n ¼ 976, P, 0.05). FIG. 3. Relationship between eggs shed in feces (real shedding) and potential shedding (eggs in utero by a population of infecting females) in single-infected rabbits for (a) T. retortaeformis and (b) G. strigosum. Fitted predictions from linear models and standard deviations (shaded area) are reported. Dashed lines depict the 1:1 ratio between the two variables.

6 June 2014 SHEDDING IN PARASITE INFECTIONS 1689 this denotes a mismatch between intensity and shedding. In other words, high G. strigosum shedding was associated with low intensities in co-infected rabbits. The amount of fecal eggs shed was comparable across the rabbit age classes (only 4þ month-old cases were used), although more eggs were shed by co-infected than single-infected hosts (Fig. 5b, n ¼ 570, P, 0.001). G. strigosum intensity of infection differed with host age (both quadratic and linear term, n ¼ 1208, P, ), and was higher in single than co-infected individuals (age quadratic term type of infection, n ¼ 1208, P, ; Fig. 5a). FIG. 4. Relationship between eggs shed in rabbit feces and parasite intensity in single (black circles) and co-infected (gray circles) rabbits for (a) T. retortaeformis and (b) G. strigosum. Fitted predictions from linear models and standard deviations (shaded areas) are reported. DISCUSSION Natural infections of a rabbit population with two gastrointestinal helminths provided new insights into the contribution of co-infections to heterogeneities in individual infectiousness. Co-infected hosts harbored higher T. retortaeformis infections, with shorter and less fecund worms, and lower G. strigosum infections, with longer and more fecund parasites. Crucially, changes in intensity of infection did not proportionally predict the observed changes in shedding: co-infected rabbits shed more G. strigosum but a similar number of T. retortaeformis eggs compared to single-infected hosts at the time of sampling. We propose that the mismatch- Female mean body length also decreased with intensity of infection (Fig. 2f, n ¼ 978, P, 0.001), although no differences were found between infection types (i.e., twoway interaction intensity infection type, P. 0.05). Among female worms, eggs in utero increased with body length consistently faster in single than co-infected hosts (Appendix: Fig. A1c, intensity type of infection, n ¼ 976, P, 0.01). Analyses repeated using the whole infrapopulation of female parasites from every rabbit confirmed these trends (Appendix: Fig. A1d, A2d, e, f). The number of G. strigosum eggs that could be potentially shed in the environment increased with parasite intensity (Fig. A3b, n ¼ 976, P, 0.001), and it was faster in co-infected than single-infected rabbits (n ¼ 976, P, 0.05). A comparison between potential shedding and real shedding, from single-infected individuals, confirmed the positive relationship between the two variables (Fig. 3b, n ¼ 90, P, 0.05, log-linear regression coefficient, 0.53), but also the lower number of eggs in the feces, as they were scaled to the standard 1 g of pellets. As previously reported for T. retortaeformis, a number of rabbits harbored female worms with a large yield of eggs in utero, but no eggs were found in the feces at the time of sampling. The relationship between eggs shed in feces and intensity of infection was then examined between the two types of infection (Fig. 4b). Real shedding was higher in co-infected than single-infected rabbits (n ¼ 608, P, 0.01) and increased with parasite intensity (n ¼ 608, P, 0.001). As highlighted for T. retortaeformis, FIG. 5. (a) Mean parasite intensity and (b) eggs shed in feces in single (black) and co-infected (gray) rabbits by host age for T. retortaeformis (circles) and G. strigosum (triangles). G. strigosum eggs are described in panel (b) right scale. Means and standard errors are reported.

7 1690 ISABELLA M. CATTADORI ET AL. Ecology, Vol. 95, No. 6 PLATE 1. Graphidium strigosum mating adults, the male holds on to the female with his caudal bursa and eggs are visible in the female s uterus. Photo credit: Nabeel Zaghtiti. ing between infection and shedding was mainly driven by changes in the fecundity intensity relationship during co-infection. While this negative association was probably the result of density-dependent parasite constraints and/or host immune-mediated regulation (Cattadori et al. 2005, 2008, Cornell et al. 2008, Murphy et al. 2011, 2013, Thakar et al. 2012), the presence of the second helminth altered these restraints and facilitated fecundity by boosting the number of eggs in utero and, ultimately, the amount of eggs shed on the herbage. However, this enhanced fecundity was not reflected in the level of infection. These findings provide empirical evidence that in addition to shaping patterns of infection (Cattadori et al. 2007, 2008, Pathak et al. 2012, Thakar et al. 2012, Murphy et al. 2013), co-infections with helminths can modulate transmission by influencing shedding, with the critical note that variation in infection might not proportionally match variation in shedding. This discrepancy is not what we expected and is an overlooked issue in epidemiological models of coinfections. By altering host susceptibility, co-infections can modify the vital rates and strategies of the incoming parasites as well as the worms already present. What determines the discrepancy between per capita eggs in utero and parasite intensity in co-infected compared to single-infected individuals is still unclear. Adjustments of fecundity and/or body length to habitat changes, such as transplantation of adult stages into naive hosts, host immunization, or host genetic makeup have been previously reported for helminths (Moqbel et al. 1980, Stear et al. 1997, Viney et al. 2006, Babayan et al. 2010). We suggest that co-infections can be an additional cause of perturbation to abundance and vital strategies, which can potentially lead to inverse density-dependent regulation of fecundity (Keymer and Slater 1987). Potential shedding, based on eggs in the female s uterus, can be used as a reliable indicator of shedding when no real shedding data (i.e., eggs in feces) are available, but needs to be confirmed. By using singleinfected rabbits, where both variables were available, we found that potential shedding was comparable to real shedding for T. retortaeformis but was higher for G. strigosum, suggesting that a fraction of eggs is lost on the way out of the host s body. This loss could be caused by rabbits being coprophagic (Thacker and Brandt 1955, Jilge 1974) and the stronger impact of this practice when the number of eggs-per-gram of pellets is relatively low, as commonly occurs for the second rather than for the first helminth species. The daily ingestion of soft feces can remove eggs from the transmission pool; these eggs are re-ingested but do not re-infect the rabbit and thus are inevitably lost, a pattern that was confirmed in our laboratory infections using this system (I. M. Cattadori, unpublished data). Co-infection does not appear to alter the coprophagic behavior, but only the fecundity intensity relationship, and thus the density of eggs shed in the environment. Since the large majority of animal

8 June 2014 SHEDDING IN PARASITE INFECTIONS 1691 species lack this habit, or adopt it irregularly, we conclude that potential shedding can be an accurate measurement of eggs shed in feces. We do not exclude that eggs can be re-absorbed by the female, although this might happen at a very early stage and not when eggs are finally formed and used for counting. Egg destruction in the gut can be another possible alternative, but we do not have any evidence to support this hypothesis. More generally, it could be argued that single and coinfected rabbits differ in their rate of shedding. Fecundity rate estimates commonly rely on the relationship between number of eggs shed in feces and female worms retrieved from the host (Keymer and Slater 1987, Anderson and May 1992). If there is a change in the rate of fecundity, it is still unclear if this is because parasites produce and release the same amount of eggs at a faster pace, or more eggs at the same pace, or a combination of both. If we assume that our snapshot of monthly sampling across the rabbit population is indicative of an average daily shedding event, by hosts of identified age and intensity of infection, our system seems to support the second mechanism. Counter to our predictions and the levels of infection observed, co-infected individuals shed more G. strigosum and a similar amount of T. retortaeformis than singleinfected rabbits. Importantly, these findings emphasize that we cannot predict the pattern of shedding in helminth co-infections based on parasite intensity alone, and vice versa. Indeed, the highly parasitized co-infected individuals might not represent the highest shedders, as a negative relationship between infection and fecundity can constrain the transmission success. While the presence of a second parasite species can turn this into a positive trend, we found that this was not enough to generate highly infected or super-shedding rabbits. This has important implications for helminth control and the selective targeting of individuals based on shedding data (i.e., epg counts), particularly if these individuals are coinfected. This observation, together with the finding that single-infected rabbits do not shed regularly despite harboring fecund worms, raises the question of how much variation in host infectiousness should be considered to correctly predict the risk of infection or the proportion of individuals to treat in order to reduce transmission below a critical threshold. As a further source of variation between individuals, we searched for the age group responsible for the majority of shedding, whether the same group carried most of the infection and whether it differed between single and co-infected hosts. Juvenile rabbits (4 and 5 months old) harbored and shed most of T. retortaeformis, while adult hosts (6þ months old) contributed the most to G. strigosum intensity and shedding. This result supports the hypothesis that for each parasite species the underlying dynamics are driven by the same host group, although level of infection and shedding varies between single and co-infected rabbits. Therefore, a change in the ratio of these two groups can be important for disease dynamics and long-term parasite persistence in the host population. In conclusion, this study represents a detailed investigation of the dynamics of infection and shedding of two gastrointestinal helminths in single and coinfected rabbits monitored monthly for eight years. Changes in host status, single or co-infected, lead to contrasting patterns of infection, fecundity, and shedding. Notably, pattern of shedding could not be predicted from intensity alone in co-infected individuals. Heterogeneities in the host response to infections have been widely recognized as one of the major drivers of disease outbreak and risk of infection. Co-infections can play a fundamental role in generating such variation. The next challenge in the ecology of infectious diseases is to explicitly consider variation in shedding in single and co-infected hosts to increase model sensitivity, and more crucially, intervention specificity. ACKNOWLEDGMENTS This study is the result of the work done by a battalion of enthusiastic undergraduate students who spent hours over four years measuring helminth specimens and antibody responses. Among the students listed as co-authors we also thank: Anna Bosset, Samantha Gaud, Han (Sol) Kan, Kristin Lambert, Anna Lojudice, Michelle Migliarese, Georgia Perry, Laken Roberts, and Maria Vigilar. This study and L. Wodzinski were supported by NSF (DEB and ). B. Wagner was supported by NSF-REU supplement funds to I. M. Cattadori. I. M. Cattadori and B. Boag also thank Douglas Allan for allowing the sampling of rabbits in his farm. The coauthors have no conflict of interest to declare. LITERATURE CITED Anderson, R. M., and R. M. May Infectious diseases of humans: dynamics and control. Oxford University Press, Oxford, UK. Babayan, S. A., A. F. Read, R. A. Lawrence, O. Bain, and J. E. Allen Filarial parasites develop faster and reproduce earlier in response to host immune effectors that determine filarial life expectancy. PLoS Biology 8:e Boag, B Helminth parasites of the wild rabbit Oryctolagus cunniculus (L.) in north east England. Journal of Helminthology 46: Burnham, K. P., and D. R. Anderson Model selection and multi-model inference: a practical information-theoretic approach. Springer, New York, New York, USA. Cattadori, I. M., R. Albert, and B. Boag Variation in host susceptibility and infectiousness generated by co-infection: the myxoma-trichostrongylus retortaeformis case in wild rabbits. Journal of the Royal Society Interface 4: Cattadori, I. M., B. Boag, O. N. Bjørnstad, S. Cornell, and P. J. Hudson Immuno-epidemiology and peak shift in a seasonal host-nematode system. Proceedings of the Royal Society B 272: Cattadori, I. M., B. Boag, and P. J. Hudson Parasite coinfection and interaction as drivers of host heterogeneity. International Journal for Parasitology 38: Chase-Topping, M., D. Gally, C. Low, L. Matthews, and M. Woolhouse Super-shedding and the link between human infection and livestock carriage of Escherichia coli O157. Nature Reviews Microbiology 6: Chylinski, C., B. Boag, M. Stear, and I. M. Cattadori Effects of host characteristics and parasite intensity on

9 1692 ISABELLA M. CATTADORI ET AL. Ecology, Vol. 95, No. 6 growth and fecundity of Trichostrongylus retortaeformis infections in rabbits. Parasitology 136: Cornell, S., O. N. Bjørnstad, I. M. Cattadori, B. Boag, and P. J. Hudson Seasonality, cohort-dependence and the development of immunity in a natural host-nematode system. Proceedings of the Royal Society B 275: Fenton, A Worms and germs: the population dynamic consequences of microparasite macroparasite co-infection. Parasitology 135: Garske, T., and C. J. Rhodes The effect of superspreading on epidemic outbreak size distributions. Journal of Theoretical Biology 253: Geerts, S., and B. Gryseels Drug resistance in human helminths: current situation and lessons from livestock. Clinical Microbiology Review 13: Gordon, H. McL., and H. Whitlock A new technique for counting nematode eggs in sheep faeces. Journal of the Council for Scientific and Industrial Research, Australia. 12: Graham, A., I. M. Cattadori, J. Lloyd-Smith, M. Ferrari, and O. N. Bjornstad Transmission consequences of coinfection: cytokines writ large? Trends in Parasitology 6: Henderson, N. G., and M. J. Stear Eosinophil and IgA responses in sheep infected with Teladorsagia circumcincta. Veterinary Immunology and Immunopathology 112: Holmes, J. C Effects of concurrent infections on Hymenolepis diminuta (Cestoda) and Moniliformis dubius (Acanthocephala). I. General effects and comparison with crowding Journal of Parasitology 88: Holmes, J. C Effects of concurrent infections on Hymenolepis diminuta (Cestoda) and Moniliformis dubius (Acanthocephala). II. Effects on Growth. Journal of Parasitology 48: Hotez, P. J Neglected infections of poverty in the United States of America. PLoS Neglected Tropical Diseases 2:e256. Hudson, P. J., A. P. Dobson, and K. D. Lafferty Is a healthy ecosystem one that is rich in parasites? Trends in Ecology and Evolution 21: Hussey, R. S., and K. R. Barker A comparison of methods of collecting inocula of Meloidogyne spp., including a new technique. Plant Disease Report 57: Jilge, B Soft feces excretion and passage time in the laboratory rabbit. Laboratory Animals 8: Kao, K. K., M. B. Gravenor, B. Charleston, J. C. Hope, M. Martin, and C. J. Howard Mycobacterium bovis shedding patterns from experimentally infected calves and the effect of concurrent infection with bovine viral diarrhoea virus. Journal of the Royal Society Interface 4: Keymer, A. E., and A. F. Slater Helminth fecundity: density dependence or statistical illusion? Parasitology Today 3: Lafferty, K. D., et al Parasites in food webs: the ultimate missing links. Ecology Letters 11: Lagrue, C., and R. Poulin Intra- and interspecific competition among helminth parasites: effects on Coitocaecum parvum life history strategy, size and fecundity. International Journal for Parasitology 38: Lass, S., P. J. Hudson, J. Thakar, J. Saric, E. Harvill, R. Albert, and S. E. Perkins Generating super-shedders: coinfection increases bacterial load and egg production of a gastrointestinal helminth. Journal of Royal Society Interface 10(80): Lloyd-Smith, J. O., S. J. Schreiber, P. E. Kopp, and W. M. Getz Superspreading and the effect of individual variation on disease emergence. Nature 438: Lustigman, S., R. K. Prichard, A. Gazzinelli, W. N. Grant, B. A. Boatinm, J. S. McCarthy, and M.-G. Basán ez A research agenda for helminth diseases of humans: the problem of helminthiases. PLoS Neglected Tropical Diseases 6:e1582. McCullers, J. A., J. L. McAuley, S. Browall, A. R. Iverson, K. L. Boyd, and H. B. Normark Influenza enhances susceptibility to natural acquisition and disease from Streptococcus pneumoniae in ferret. Journal of Infectious Diseases 202: Moqbel, R., D. J. Mclaren, and D. Wakelin Strongyloides ratti: reversibility of immune damage to adult worms. Experimental Parasitology 49: Murphy, L., N. Nalpas, M. Stear, and I. M. Cattadori Explaining patterns of infection in free living populations using laboratory immune experiments. Parasite Immunology 33: Murphy, L., A. K. Pathak, and I. M. Cattadori A coinfection with two gastrointestinal nematodes alters host immune responses and only partially parasite dynamics. Parasite Immunology. Over, J. H., J. Jansen, and P. W. van Olm Distribution and impact of helminth diseases of livestock in developing countries. FAO Animal Production and Health Paper 96. FAO, Rome, Italy. Paterson, S., and R. Barber Experimental evolution of parasite life-history traits in Strongyloides ratti (Nematoda). Proceedings of the Royal Society B 274: Pathak, A. K., C. Pelenski, B. Boag, and I. M. Cattadori Immuno-epidemiology of chronic bacteria-helminth co-infections: observations from the field and evidences from the laboratory. International Journal for Parasitology 4: Pinheiro, J. C., and D. M. Bates Mixed-effects models in S and S-PLUS. Springer, New York, New York, USA. Quinnell, R. J., G. F. Medley, and A. E. Keymer The regulation of gastrointestinal helminth populations. Proceedings of the Royal Society B 330: Stear, M. J., K. Bairden, J. L. Duncan, P. H. Holmes, Q. A. McKellar, M. Park, S. A. J. Strain, and M. Murray How hosts control worms. Nature 389:27. Stear, M. J., S. C. Bishop, M. Doligalska, J. L. Duncan, P. H. Holmes, J. Irvine, L. Q. A. McKellar, E. Sinski, and M. Murray Regulation of egg production, worm burden, worm length and worm fecundity by host responses in sheep infected with Ostertagia circumcincta. Parasite Immunology 17: Taylor, M. A., R. L. Coop, and R. L. Wall Veterinary parasitology. Third edition. Blackwell, Oxford, UK. Thacker, E. J., and C. S. Brandt Coprophagy in the rabbit. Journal of Nutrition 55: Thakar, J., A. K. Pathak, L. Murphy, R. Albert, and I. M. Cattadori Network model of immune responses reveals key effectors to single and co-infection kinetics by a respiratory bacterium and a gastrointestinal helminth. PLoS Computational Biology 8:e Viney, M. E., M. D. Steer, and C. P. Wilkes The reversibility of constraints on size and fecundity in the parasitic nematode Strongyloides ratti. Parasitology 133: SUPPLEMENTAL MATERIAL Appendix Detailed description of antibody preparation and detection, and additional figures describing the relationships between parasite characteristics in single and co-infected rabbits (Ecological Archives E A1).