INTRODUCTION
There has been much recent interest into how complex life cycle parasites optimize their larval growth (Choisy et al. Reference Choisy, Brown, Lafferty and Thomas2003; Parker et al. Reference Parker, Chubb, Ball and Roberts2003a,Reference Parker, Chubb, Roberts, Michaud and Milinskib; Duclos et al. Reference Duclos, Danner and Nickol2006; Iwasa and Wada, Reference Iwasa and Wada2006; Michaud et al. Reference Michaud, Milinski, Parker and Chubb2006; Ball et al. Reference Ball, Parker and Chubb2008; Shostak et al. Reference Shostak, Walsh and Wong2008; Parker et al. Reference Parker, Ball and Chubb2009a,Reference Parker, Ball and Chubbb; Benesh, Reference Benesh2010a,Reference Beneshb; Chubb et al. Reference Chubb, Ball and Parker2010). On the one hand, a fast larval growth rate and a large size at infectivity are expected to increase fitness. Rapid growth reduces the time until developmental thresholds are reached, thus lowering the chances of dying before being capable of transmission to the next host (Day and Rowe, Reference Day and Rowe2002), whereas a large transitional size may increase infection success (Steinauer and Nickol, Reference Steinauer and Nickol2003) or adult fecundity (Fredensborg and Poulin, Reference Fredensborg and Poulin2005). On the other hand, rapid growth entails an elevated consumption of host resources, and may thus decrease host survival and parasite transmission. This trade-off between growth and transmission is thought to shape larval life history (Parker et al. Reference Parker, Chubb, Roberts, Michaud and Milinski2003b).
In the case of larval helminths, normal infection levels are very low (frequently a single worm per host) (Poulin, Reference Poulin2007), and infection does not always lower host survival (Poulin et al. Reference Poulin, Curtis and Rau1992; Hurd et al. Reference Hurd, Warr and Polwart2001; Guinnee and Moore, Reference Guinnee and Moore2004; Benesh, Reference Benesh2010b). This suggests that some parasites may not normally exploit their hosts at levels where mortality costs are measurable. In such systems, mortality costs may only be detected at unnaturally high levels of parasite growth. The rate and amount of parasite growth can be manipulated by experimentally producing a range of infection intensities. The total biomass accumulated by multiple worms should exceed that of a single worm, unless single parasites are maximally exploiting the available host resources (Parker et al. Reference Parker, Chubb, Roberts, Michaud and Milinski2003b). Thus, if there is a trade-off between total worm growth and host viability, then we expect some pattern of intensity-dependent mortality (IDM).
Older studies of IDM in larval helminths were stimulated by population models (Anderson and May, Reference Anderson and May1978), and focused on whether parasites regulated intermediate host populations (Keymer, Reference Keymer1980; Uznanski and Nickol, Reference Uznanski and Nickol1980; Skorping, Reference Skorping1984, Reference Skorping1985). Here, I emphasize the insights into larval growth strategies that may be gleaned from IDM studies. In particular, the relationship between total parasite biomass and intensity indicates the infection levels at which host resources limit larval growth, whereas patterns of IDM indicate when and if ‘host mortality’ costs are manifested. I start by outlining a simple conceptual framework in which to categorize patterns of IDM. Using this framework, the patterns of intensity-dependent helminth growth and intermediate host survival found in the literature are summarized. An example study of IDM using the nematode Camallanus lacustris and its copepod intermediate host is also provided. The aims in this paper are to (1) evaluate the typical relationship between larval parasite growth/size and infection intensity, (2) assess the patterns of IDM associated with parasite growth, and (3) establish whether normal levels of intermediate host exploitation entail risks for parasites.
MATERIALS AND METHODS
Conceptual background for categorizing the literature
Complex life cycle parasites use resources from their intermediate hosts to fuel their larval growth. Therefore, by quantifying overall parasite growth (how fast and how much), we can estimate the impact that parasites have on the host. However, if worms vary considerably in their biochemical makeup, i.e. some are of higher quality than others, then worm size need not be well correlated with the actual drain on host resources. Indeed, the effect of helminth infection on the intermediate host can depend on the availability of particular macronutrients (Ponton et al. Reference Ponton, Lalubin, Fromont, Wilson, Behm and Simpson2011). Nonetheless, overall parasite biomass typically varies several-fold with intensity, and this extensive variation in parasite size should be at least roughly related to the consumption of host nutrients.
The generalized relationship between intensity and burden imposed on the host is likely to be an asymptotic function (Fig. 1A) (Parker et al. Reference Parker, Chubb, Roberts, Michaud and Milinski2003b). Detriment to the host initially increases as more parasites sap host nutrients, but this should eventually level off as parasites approach the maximum possible level of host exploitation. By estimating the shape of this function over a natural and unnatural range of intensities, we can assess how extensively a parasite species normally exploits its intermediate host. Consider 2 extremes. When the burden on the host increases proportionally over a large range of intensities, i.e. many individuals can infect a host without approaching a maximum level of exploitation, this suggests the parasite is a rather mild-exploiter of the host. An example of such parasites might be acanthocephalans in crabs (Poulin et al. Reference Poulin, Nichol and Latham2003). On the other end of the spectrum, the resources consumed and growth attained by a parasite may already be maximized in single-worm infections. A potential example of such aggressive parasites might be some tapeworms infecting copepods (Parker et al. Reference Parker, Chubb, Roberts, Michaud and Milinski2003b).
Fig. 1. (A) Generalized relationship between intensity of infection (arbitrary units) and impact on the host, which can presumably be indirectly assessed by quantifying parasite growth. Hypothetical threshold burdens, beyond which host survival decreases, are indicated by dashed lines labelled I to V. (B and C) The predicted patterns of intensity-dependent mortality are shown for a parasite exhibiting a decelerating intensity-burden relationship (dark grey box) as well for the case in which burden increases linearly with intensity (light grey box). The survival patterns are shown for the particular ‘danger’ thresholds I–V.
Two areas of the function in Fig. 1A are highlighted, the initial, near-linear trajectory (light grey box) and the non-linear, decelerating advance to the maximum (dark grey box). Assume that we observe these two intensity-burden relationships. What are the expected patterns of IDM? This depends on the parasite burden that hosts can tolerate. The dashed lines in Fig. 1A indicate hypothetical burdens beyond which IDM occurs. In the parlance of recent theoretical models, these thresholds represent the total parasite mass at which the noise mortality rate for the parasite becomes size-dependent (Parker et al. Reference Parker, Chubb, Roberts, Michaud and Milinski2003b; Ball et al. Reference Ball, Parker and Chubb2008; Parker et al. Reference Parker, Ball and Chubb2009a). Different thresholds produce different qualitative patterns of IDM (Fig. 1B and C). If hosts are sensitive to even low parasite burdens, then each increase in intensity should decrease host survival, either proportionally (threshold I, Fig. 1C) or with a slowing rate (threshold III, Fig. 1B). Alternatively, hosts may be able to harbour low numbers of parasites with no ill effect, but once a threshold is passed, survival decreases, resulting in a discontinuous pattern of IDM (threshold II in Fig. 1C and threshold IV in Fig. 1B). When parasites are far from maximally exploiting the host, even a steep impact-intensity relationship could be tolerable, so no IDM is expected (thresholds III to V, Fig. 1C). At the exploitation maximum, increases in intensity do not add much burden to the already strained host. If hosts are able to tolerate this extreme, then IDM is never expected (threshold V, Fig. 1B).
Survey and categorization of the literature
What is the typical relationship between intensity and growth? And is mortality related to growth? To address these questions, I searched for studies on intensity-dependent growth and mortality in the larval stages of trophically transmitted helminths. I focused on species that do not asexually reproduce in their intermediate host. Studies were found during a more comprehensive survey of growth and development in larval helminths (Benesh et al. 2011), in which literature databases (Web of Science, PubMed, Google Scholar) were queried with genus names and the terms ‘life cycle’, ‘growth’, or ‘development’.
Two growth characteristics may affect host wellbeing, rate and final size attained. Typically, investigators have recorded how average parasite size, but not growth rate, changes with intensity. These two parameters are likely correlated, but it should be kept in mind that they do not necessarily produce the same intensity-burden relationships. I was interested in how overall parasite growth affects host viability, so I examined how the total parasite burden (size) scaled with intensity. Species were categorized into 1 of 3 intensity-burden relationships. When each additional worm increased the total by about the same amount, then burden was considered proportional to intensity (the light grey area in Fig. 1A). Note that a proportional increase in total worm size with intensity does not imply the absence of crowding effects. When the slope of the intensity-burden relationship is less than 1 the average worm size decreases with intensity. If the total increased by a smaller and smaller amount with each worm, then the burden-intensity relationship was categorized as decreasing (the dark grey area in Fig. 1A). Finally, if the maximum burden has been reached already at the lowest observed intensities, then the total worm biomass may not increase at all with intensity.
For intensity-mortality relationships, the hypothetical scenarios of IDM depicted in Fig. 1B and 1C were used to categorize observed patterns. Potential relationships between survival probability and intensity were: proportional (case I), flat and then a linear decrease (case II), decreasing with a slowing rate (III), flat followed by a non-linear decrease (IV), or completely flat (V). Unfortunately, the ability to discriminate between case II and IV is low, because in most studies, there has been little replication within intensity levels and the range of observed intensities may not be large enough to differentiate a non-linear from a linear decrease. Thus, cases in which host survival decreased past a certain intensity level were simply considered to have a discontinuous pattern of IDM.
An example: the effect of intensity on growth and mortality in Camallanus lacustris
The nematode Camallanus lacustris has a 2-host life cycle. Free, first-stage larvae are eaten by freshwater copepods where they invade the body cavity. The worms undergo 2 moults in copepods before reaching the infective L3 stage after about 12 days at room temperature (Moravec, Reference Moravec1969). Several species of fish are potential final hosts, but perch (Perca fluviatilis) is probably the most frequent. As nematodes must reach the L3 stage before being infective (Anderson, Reference Anderson2000), the final size attained by larvae was expected to be relatively constant, at least in comparison to cestodes or acanthocephalans. Consequently, unless growth rates varied considerably, the total burden on the host was predicted to increase proportionally with intensity.
Experimental infection and maintenance of copepods
Gravid female worms were collected from the guts of perch caught in the Grosser Plöner See, Germany (54º09′N 10º25′E). L1 larvae were dissected from 10 female worms, pooled together, and then kept at 4°C overnight before being used for infection the following day. Copepods (Macrocyclops albidus) were taken from a laboratory culture (see van der Veen and Kurtz (Reference Van der Veen and Kurtz2002)). Two groups of copepods were used for infection; adult males and adult females. Female copepods are larger than males, and likely provide more nutrients and/or space to developing larvae. Copepods were isolated individually in 24-well microtitre plates (n=930). To create a range of infection intensities, copepods were exposed to 2, 4, or 6 C. lacustris larvae 2 days after isolation. Some copepods served as unexposed controls (n=111). Copepods were maintained at 18°C with a 18:6 h light:dark cycle, and they were fed every second day with 3 freshly hatched Artemia salina nauplii.
Copepod dissection and larval parasite growth
Copepods were checked daily and dead individuals were frozen at −20°C. Dead copepods were later thawed and dissected to determine the intensity of infection. Copepods that died within the first 2 days post-exposure were not processed (n=28), because the L1 larvae can actively leave the dead host, leading to under-estimates of intensity. Additionally, to quantify worm growth rates, random subsamples of live copepods were taken 5 and 11 days post-exposure (n=164 and n=160, respectively). On these days, worms are typically undergoing the L1 to L2 moult and the L2 to L3 moult (Moravec, Reference Moravec1969). After 75 days, the experiment was terminated and all remaining copepods were frozen.
For most of the infected copepods, all worms were measured. Worms were placed on a microscope slide in 10 μl of water under a cover-slip. Each worm was photographed with a digital camera, and length and width were measured using the freeware Image J 1·38×(Rasband, W.S., NIH, Bethesda, Maryland, USA, http://rsb.info.nih.gov/ij/, 1997–2009). Body width was recorded where the worm's pharynx ended. Worms were considered cylindrical in shape, so their volume (mm3) was estimated with the equation (πlw 2)/4 where l is worm length and w is worm width. The weight of the cover-slip compressed worms slightly. This was necessary, because L3 worms are often coiled, making their length otherwise difficult to measure. However, because of this flattening the worm volumes presented here may be somewhat distorted in comparison with other studies (Moravec, Reference Moravec1969).
Relationship between intensity and burden on the host
To assess how growth rate and final size change with intensity, growth curves were fitted to the data for each intensity level. The following asymptotic function was used: V(t)=2·06×10−4 + A × (eB/t) where V is the total worm volume harboured by a copepod, A is the asymptotic volume, and B is the relative rate at which the asymptote is approached (see Shostak et al. (Reference Shostak, Rosen and Dick1985) and Michaud et al. (Reference Michaud, Milinski, Parker and Chubb2006) for similar approaches). The function's intercept was set at 2·06×10−4 mm3 because this was the average volume of free-living infective L1 larvae (n=23). The parameters for this non-linear function, as well as their standard errors, were estimated via an iterative procedure executed with SPSS version 18.0 (SPSS Inc., Chicago, Illinois, USA). Parasite growth was analysed separately in male and female copepods.
Relationship between intensity and host survival
The effect of intensity on copepod survival was assessed with Cox regression, a method commonly used for survival analyses (Andersen, Reference Andersen1991). Male and female copepods had clearly different life spans (average survival of males was 19·7 days versus 54·9 days for females), so they were analysed separately. Uninfected copepods were either unexposed controls or exposed, but uninfected. These 2 groups did not differ significantly in a preliminary Cox regression, so for simplicity all uninfected copepods were pooled for the analysis. Intensity was entered into the regression model as a categorical covariate. An assumption of Cox regression models is that the ratio of the hazard function for any 2 individuals is dependent on their covariate values and the baseline hazard function, but not time (proportional hazards assumption). However, I was interested in whether intensity-dependent mortality occurs at particular times, i.e. during or after the major growth phase, so I checked the validity of the proportional hazards assumption. Two predictor terms were entered into a preliminary model: intensity alone as well as the interaction between intensity and a time-dependent covariate. If the intensity by time interaction explained more variation (had a higher Wald χ 2 value), then the effect of intensity was considered to be time dependent and intensity alone was removed from the model.
Checking dose effects
Copepods harbouring the same number of worms may not be entirely comparable, because they were exposed to different doses. For example, a copepod with 1 worm, depending on the dose it received, may have been infected by 50%, 25% or 12·5% of the larvae to which it was exposed. If susceptibility is correlated with host quality, then within each intensity level dose may affect parasite growth or host survival. To check this, the data were split by copepod sex and by intensity level, so that within groups only dose varied. For each group, a Cox regression was performed with dose as the only predictor. ANOVAs were also conducted using these data subsets to assess whether total worm volume depended on dose. As worms were measured at different time-points, the values for worm size were ‘corrected’ for time by taking the residuals of the fitted growth curves (e.g. see Fig. 2). These residuals were then used as dependent variables in the ANOVAs.
Fig. 2. The total volume (mm3) of larval Camallanus lacustris at different infection intensities in (A) female copepods and (B) male copepods.
RESULTS
Intensity-dependent growth and mortality in larval Camallanus lacustris
Dose effects
Sample sizes were large enough to confidently assess dose effects for only intensities of 1, 2, and 3. For both male and female copepods, dose did not significantly affect copepod survival at any infection level (Cox regressions, all P>0·089). Likewise, there were no clear effects of dose on parasite growth (ANOVAs, all F<2·154, P>0·101). Consequently, in the following analyses intensity was considered without regard to dose.
Larval growth at different intensities
There were not many male copepods infected with 5 or 6 larvae (n=10 and 14, respectively), so growth curves were only fitted for the intensity levels 1–4. In general, the growth curves fit the data well, usually explaining over 70% of the variation in total worm volume (Table 1, Fig. 2). Parasite growth differed between the host sexes; the estimates for asymptotic total worm volume and relative growth rate tended to be lower in male copepods (Table 1). The asymptotic size increased with intensity, particularly in female copepods (Fig. 3), but within each copepod sex, growth rate estimates were rather consistent, indicating that the asymptotic volume for each intensity level is approached at about the same relative rate. Additionally, there was a weak crowding effect. In single infections, the asymptotic volume was about 0·003 mm3, but the total worm volume did not increase by 0·003 mm3 with each additional worm (Fig. 3). In summary, total volume increases proportionally with intensity, albeit at a faster rate in female copepods. In male copepods, worm volume seems to level off above an intensity of 3 worms (Fig. 3), suggesting that the slope of the intensity-burden relationship might be decreasing past this intensity.
Fig. 3. Estimated asymptotic total volume (mm3) of Camallanus lacustris larvae reached at different infection intensities in female and male copepods. The diagonal line depicts the expected increase in worm volume without crowding effects, i.e. when worms from multiple infections are the same size as those from single infections. Bars represent s.e.
Table 1. Parameter estimates for the fitted growth curves
Host survival
For female copepods, but not males, the intensity by time interaction had a higher Wald χ 2 score than intensity alone in a preliminary Cox regression analysis. Thus, for males just intensity was a predictor in the Cox regression, whereas for females the intensity by time interaction was checked. Intensity did not affect the survival of male copepods (Wald χ 2=5·941, d.f.=4, P=0·204). For female copepods, however, the intensity by time interaction was significant, with copepods harbouring 4 or more worms having slightly lower survival (Table 2). Pronounced mortality in female copepods was only observed after about 20 days, with individuals harbouring 4–6 worms dying at an elevated rate after this time (Fig. 4).
Fig. 4. Survival of female copepods infected with 1 to 6 Camallanus lacustris larvae. The black solid line represents the survival of uninfected copepods (both unexposed controls and exposed, but uninfected individuals).
Table 2. Results of Cox regression survival analysis assessing intensity-dependent mortality in female copepods (N=288)
Summarizing the literature
Studies fell into 3 categories: (1) only intensity-dependent growth recorded, (2) only intensity-dependent mortality described, or (3) both. Studies of the second type are difficult to interpret, because mortality cannot be correlated with growth, so they are not discussed further here (Kisielewska, Reference Kisielewska1959; Huizinga, Reference Huizinga1967; Rosen et al. Reference Rosen, Ash and Wallace1970; Moravec, Reference Moravec1978; Courtney and Christensen, Reference Courtney and Christensen1987; Okaka, Reference Okaka1989; Solomon et al. Reference Solomon, Paperna and Alkon1996; Lopez et al. Reference Lopez, Panadero, Diez and Morrondo1998; Meissner and Bick, Reference Meissner and Bick1999; Ferreira et al. Reference Ferreira, Jensen, Martins, Sousa, Marques and Pardal2005; Hansen and Poulin, Reference Hansen and Poulin2005; Kokkotis and McLaughlin, Reference Kokkotis and McLaughlin2006). Moreover, a number of studies could not be categorized into the outlined framework. For instance, a few studies compared just 2 intensity levels (high vs low), making it impossible to evaluate whether the relationship between intensity and growth/mortality is linear or non-linear (Measures, Reference Measures1988; Robert and Gabrion, Reference Robert and Gabrion1991; Sandland and Goater, Reference Sandland and Goater2000; Dezfuli et al. Reference Dezfuli, Giari and Poulin2001). Also, some authors wrote that larval growth was intensity-dependent, but did not support these observations with data (Calentine, Reference Calentine1965; Awachie, Reference Awachie1966; Denny, Reference Denny1969; Wootten, Reference Wootten1974; Korting, Reference Korting1975).
My main goal was to evaluate the connection between total parasite growth and intermediate host mortality, and I found 10 species in which intensity, growth, and host mortality were quantified (Table 3). Nine of these cases involved helminths in their first host, while only 1 focused on a trematode exploiting its second intermediate host (Fredensborg et al. Reference Fredensborg, Mouritsen and Poulin2004). I found an additional 23 cases in which only intensity-dependent growth was recorded (Table 4). Most studies involved experimental infections (85%) that produced intensities far exceeding those typically observed in nature. In every study, parasite impact on the host, usually gauged by the total worm size, increased with intensity (Tables 3 and 4). However, in 21% of cases the increase had a decreasing slope, suggesting that the burden on the host often approaches a maximum level at higher intensities.
Table 3. Studies on complex life cycle helminths that have examined the relationship between intensity, larval parasite growth and intermediate host mortality
* (C), cestode; (A), acanthocephalan; (N), nematode; (T), trematode.
Table 4. Studies on complex life cycle helminths that reported the relationship between intensity and some aspect of larval growth, typically the final size
* (C), cestode; (A), acanthocephalan; (N), nematode; (T), trematode.
Occasionally, patterns of IDM predictably follow patterns of intensity-dependent parasite growth. For example, the total larval growth rate of the nematode Elaphostrongylus rangiferi increases linearly with intensity, and accordingly snail survival rate decreases linearly with intensity (Skorping, Reference Skorping1984, Reference Skorping1985). However, there are several studies in which IDM is discontinuous and thus less predictable (Table 3). Three of the 5 cases of discontinuous IDM occurred when the total burden on the host appeared to approach a maximum, suggesting that host exploitation only decreases host survival at very high levels (Rosen and Dick, Reference Rosen and Dick1983; Duclos et al. Reference Duclos, Danner and Nickol2006; Michaud et al. Reference Michaud, Milinski, Parker and Chubb2006).
The absence of IDM has been observed under very different conditions. On the one hand, Sakanari and Moser (Reference Sakanari and Moser1985) observed lowered survival in infected copepods, but this did not depend on the number of tapeworm larvae they harboured. At the very high intensities in their study, the exploitation of the host was maximized, suggesting that extreme host exploitation entails risks but that it does not matter how many individual parasites are actually exploiting the host. On the other hand, Uznanski and Nickol (Reference Uznanski and Nickol1980) found no IDM in amphipods infected with the acanthocephalan Leptorhynchoides thecatus, although the total parasite volume increases proportionally with intensity, at least in the studied intensity range (see Steinauer and Nickol, Reference Steinauer and Nickol2003 for the intensity-parasite size relationship). This suggests that these worms are far from maximally exploiting their hosts and that amphipods can tolerate intensities above the normal range.
Finally, 1 study does not fit into any of the scenarios presented in Fig. 1. Working with the tapeworm Hymenolepis diminuta, Keymer (Reference Keymer1980) observed a linear decrease in host survival with intensity, even though total worm size increased with a decelerating slope with intensity. This suggests that parasite growth is not the cause of mortality. In this case, as well as in the lone trematode example (Fredensborg et al. Reference Fredensborg, Mouritsen and Poulin2004), other mechanisms that are proportional to intensity are presumably responsible for host death, such as the damage done to the host during invasion of the body cavity.
DISCUSSION
Helminths that aggressively drain host resources to fuel their own growth may increase the chances of host death and thus failed transmission. This trade-off between using intermediate hosts for nutrition and transportation is thought to shape parasite life-history strategies (Choisy et al. Reference Choisy, Brown, Lafferty and Thomas2003; Parker et al. Reference Parker, Chubb, Roberts, Michaud and Milinski2003b; Iwasa and Wada, Reference Iwasa and Wada2006; Ball et al. Reference Ball, Parker and Chubb2008; Parker et al. Reference Parker, Ball and Chubb2009a; Chubb et al. Reference Chubb, Ball and Parker2010). Experimental manipulation of the burden on the host by using a range of infection intensities is one way to evaluate this trade-off, and I have summarized the results from such studies. Only 15% of the cases in Tables 3 and 4 involve the second intermediate host; most studies to date focus on the first host. Helminths usually enter the first intermediate host as eggs or small larvae before rapidly increasing in size, so growth costs and IDM are intuitively expected. Size at establishment in the second intermediate host is much larger, which likely favours reduced growth or paratenesis (Poulin and Latham, Reference Poulin and Latham2003; Parker et al. Reference Parker, Ball and Chubb2009a; Chubb et al. Reference Chubb, Ball and Parker2010). Perhaps researchers have assumed that the relatively low levels of growth in second intermediate hosts are unlikely to affect host viability, explaining the fewer studies at this life-cycle stage.
What is the typical growth-intensity relationship and what does it imply?
In all studies to date, the overall parasite mass increased with intensity, but in some species, this increase decelerated at high intensities. These observations indicate that (1) at low, natural intensities individual parasites do not exploit the host at a maximum level, and (2) resource ceilings are detectable and some species may approach them sooner than others. When resource ceilings have been observed, it has been at either extreme intensities (Pilecka-Rapacz, Reference Pilecka-Rapacz1986; Heins et al. Reference Heins, Baker and Martin2002; Benesh and Valtonen, Reference Benesh and Valtonen2007) or in small hosts (Rosen and Dick, Reference Rosen and Dick1983; Sakanari and Moser, Reference Sakanari and Moser1985; Michaud et al. Reference Michaud, Milinski, Parker and Chubb2006). Indeed, in C. lacustris there were hints of a resource ceiling in male copepods but not in larger female copepods. Even here, though, indications of a ceiling only appear at an intensity of 3, which is probably 3 times higher than the normal infection level. Thus, at low intensities larval growth is not obviously constrained by resources.
Why don't larval worms take advantage of all resources available? One possibility is that the benefits associated with larger size are marginal. For instance, if worms are able to grow faster with lower mortality in the next host, then there may be little incentive to spend a long time extensively exploiting the first host (Werner and Gilliam, Reference Werner and Gilliam1984; Abrams et al. Reference Abrams, Leimar, Nylin and Wiklund1996; Choisy et al. Reference Choisy, Brown, Lafferty and Thomas2003; Parker et al. Reference Parker, Chubb, Ball and Roberts2003a; Gandon, Reference Gandon2004; Iwasa and Wada, Reference Iwasa and Wada2006; Ball et al. Reference Ball, Parker and Chubb2008; Parker et al. Reference Parker, Ball and Chubb2009a). Additionally, the exact relationship between a large larval size and its potential advantages, like increased infectivity (Rosen and Dick, Reference Rosen and Dick1983; Steinauer and Nickol, Reference Steinauer and Nickol2003), is not well explored. Perhaps extremely large worms reap few additional fitness benefits. The benefits associated with rapid growth seem less disputable; the faster worms reach a minimum size for infectivity, the less likely they will die before transmission (Day and Rowe, Reference Day and Rowe2002). Although overall parasite biomass tends to accumulate faster with increasing intensity, this does not necessarily imply that individual parasites could grow faster. Individual growth might be limited by some intrinsic factor, like the speed of nutrient uptake, rather than overall resource availability. However, under normal circumstances most organisms do not grow at a maximum rate (Gotthard, Reference Gotthard and Thorndyke2001; Metcalfe and Monaghan, Reference Metcalfe and Monaghan2001), and the substantial phenotypic plasticity and considerable variation across species hints that larval helminth growth may be rather labile (Shostak et al. Reference Shostak, Walsh and Wong2008; Benesh, Reference Benesh2010a). Despite these alternatives, the most frequently cited limitation on parasite growth is virulence costs.
Does growth determine IDM? And are natural levels of intermediate host exploitation risky?
As total parasite growth usually increases with intensity, we expect host mortality to be intensity dependent, assuming there is a trade-off between growth and mortality. In 2 cases, host mortality has indeed been observed to increase proportionally with intensity (Skorping, Reference Skorping1984, Reference Skorping1985; Nie and Kennedy, Reference Nie and Kennedy1993). Surprisingly, though, a number of studies found discontinuous IDM, suggesting only extreme parasite burdens reduce host survival. And some studies have failed to find any IDM, such as the case study presented here. Presumably, all C. lacustris larvae must reach a threshold size to moult to the L3 stage. As all worms need to achieve some minimum amount of growth, virulence-reducing plastic responses to intensity are perhaps less likely, such as smaller final sizes (Parker et al. Reference Parker, Chubb, Roberts, Michaud and Milinski2003b). Although final size appears somewhat flexible, given the different sizes attained in male and female hosts, the burden on the host increased proportionally with intensity. However, no IDM was observed in male copepods and in females IDM was only observed at the highest intensities. Thus, copepods were able to tolerate all but the most extreme parasite burdens. This result is not unique; the literature summary indicates that intermediate hosts can often survive exceptional infection levels. This implies that at low, natural intensities, larval growth is often not at a level that induces any host mortality (Uznanski and Nickol, Reference Uznanski and Nickol1980; Poulin et al. Reference Poulin, Curtis and Rau1992; Hurd et al. Reference Hurd, Warr and Polwart2001; Guinnee and Moore, Reference Guinnee and Moore2004; Benesh, Reference Benesh2010b).
In female copepods infected with C. lacustris, IDM was time dependent. Decreased survival in hosts with 4 or more worms occurred mainly after 20 days post-exposure (dpe), when worm growth had nearly slowed to a stop. This is surprising, as host exploitation presumably slows with the arrest of worm growth, and it suggests that the mortality associated with excessive growth may be delayed. The vast majority of worms reached the L3 stage by 11 dpe (data not shown, but in accordance with Moravec (Reference Moravec1969)), so most appear to be capable of transmission before any increase in mortality. Thus, growth costs could be completely avoided if transmission occurs quickly enough. Day (Reference Day2003) argued that if virulence costs (e.g. a decrease in transmission due to host mortality) usually occur after virulence benefits have been accrued (e.g. a large larval size), then selection should favour increased virulence, because there is only a low probability that its costs will be paid. Parker et al. (Reference Parker, Ball and Chubb2009a) assumed that larval growth arrest serves to reduce host mortality. If mortality actually increases after arrest, as observed here, then there is no clear reason for parasites to stop growing, and a highly aggressive growth strategy may be optimal (Parker et al. Reference Parker, Ball and Chubb2009a). The delayed mortality of copepods heavily infected with C. lacustris further suggests that individual parasites are much less aggressive that they could be and perhaps less than what is theoretically favourable. Future studies should also note the temporal pattern of IDM, given the important implications for virulence evolution (Chubb et al. Reference Chubb, Ball and Parker2010).
To conclude, overall parasite growth/size generally increases with intensity, but frequently this does not result in a proportional increase in host mortality. Indeed, IDM is often absent or only observed at extreme levels. This indicates that (1) at natural intensity levels parasites do not exploit the host as much as they presumably could and that (2) increased growth would not entail obvious mortality costs. It is important to note that all these studies have been conducted in rather benign laboratory conditions, i.e. ad libitum food, absence of predation and competition. Under less favourable conditions, larval parasite growth may have a much larger impact on host survival. A number of field surveys have uncovered patterns indicative of IDM in larval helminths (Crofton, Reference Crofton1971; Amin et al. Reference Amin, Burns and Redlin1980; Brattey, Reference Brattey1986; Thomas et al. Reference Thomas, Renaud, Rousset, Cezilly and Demeeus1995; Brown et al. Reference Brown, Loot, Grenfell and Guegan2001; Latham and Poulin, Reference Latham and Poulin2002; Brown et al. Reference Brown, De Lorgeril, Joly and Thomas2003; Outreman et al. Reference Outreman, Cezilly and Bollache2007; Bates et al. Reference Bates, Poulin and Lamare2010; Heins et al. Reference Heins, Birden and Baker2010), but the source of host mortality cannot usually be unambiguously inferred (e.g. transmission to the next host or just a generally higher death rate?), so the impact on the costs and benefits of larval parasite growth is unclear. Undoubtedly, additional experiments incorporating natural causes of mortality, such as food limitation or predation, are needed before concluding that mortality is not an important constraint on larval helminth growth.
ACKNOWLEDGEMENTS
Thanks to Bettina Bartels, Monika Wulf, and Regina Leipnitz for help maintaining copepods. I am also grateful to two anonymous reviewers for their helpful comments.
