Parasite in uences on host life history: Echinostoma revolutum parasitism of Lymnaea elodes snails

Transcription

1 Oecologia (1998) 115:188±195 Ó Springer-Verlag 1998 Robert E. Sorensen á Dennis J. Minchella Parasite in uences on host life history: Echinostoma revolutum parasitism of Lymnaea elodes snails Received: 5 September 1997 / Accepted: 19 November 1997 Abstract Using eld surveys and experimental infections, we investigated the in uence of a trematode parasite on life history traits of adult Lymnaea elodes snails. We found that parasitism signi cantly a ected the growth, fecundity, and survival of host snails. Within ve of the six natural L. elodes populations we sampled, shell length of echinostome-infected hosts was signi cantly greater than for uninfected conspeci cs. Furthermore, we show that gigantism occurs among experimentally infected snails due to an accelerated growth rate and size-selective mortality following an Echinostoma revolutum infection. The fecundity of infected snails sharply decreased beginning at 3 weeks post exposure (PE) and all egg production eventually ceased for most hosts by 5±6 weeks PE. Energy constraints, imposed by parasite development, alter the host energy budget. Early in the infection, parasite depletion of host energy reserves reduces host reproduction, but su cient resources remain to allow accelerated host growth. Mortality was increased among host snails at two distinct stages: shortly after exposure and several weeks after cercariae were rst released. We did not observe tissue degradation in snails during the rst 4 weeks after exposure to the parasite, but destruction of host tissues was noted among snails dying later in the infection. Key words Host life history á Parasitism á Gigantism á Echinostoma á Lymnaea Introduction An organism's di erential investment into growth, reproduction, and survival de nes its life history strategy. Allocation trade-o s exist between these processes such R.E. Sorensen (&) á D.J. Minchella Department of Biological Sciences, Purdue University, West Lafayette, IN 47907, USA sorensen@bilbo.bio.purdue.edu; Fax: that energy channeled to any one of them is unavailable to the remaining two components (Stearns 1976, 1992). The lifetime reproductive success of an organism is a manifestation of the life history traits expressed by that organism throughout its life. Interspeci c interactions, such as predation (Crowl and Covich 1990) and competition (Brown 1982), in uence life history traits; likewise, parasitism shapes individual host life history strategies (Sousa 1983; Minchella 1985). Price (1980) elegantly emphasized the potential relationship between parasitism and host life history strategies, but researchers have just begun to elucidate this interaction during the last decade (Esch and Fernandez 1994; Michalakis and Hochberg 1994). By utilizing eld collections and experimental trematode infections of Lymnaea elodes, a common freshwater snail, we assessed parasite in uences on host growth, fecundity, and survival. Parasites impose direct tness costs on their hosts by stealing nutrients normally used for host life processes (Toft 1991). Trematodes are a diverse group of endoparasites requiring at least two hosts: a mollusc and a vertebrate. Intramolluscan trematode parasitism is frequently associated with the alteration of a host's growth, fecundity, or survival; yet, few clear patterns exist (reviewed in Thompson 1997). The results vary depending upon several factors, including: (1) the speci c hostparasite combination (Minchella and LoVerde 1983; Schrag and Rollinson 1994), (2) host age (Loker 1979; Kuris 1980; Keas and Esch 1997), (3) the life span and typical reproductive pattern of the host (Sousa 1983; Curtis 1995), and (4) whether the experiments were conducted in a laboratory or in the eld (McClelland and Bourns 1969; Fernandez and Esch 1991). However, in the midst of these host response variations, there appears to be one common outcome among molluscs following trematode infection: an eventual elimination of host reproductive output, often termed ``parasitic castration'' (Wright 1971; Malek and Cheng 1974). For uninfected freshwater snails, the production of energetically expensive egg masses occurs at the expense of body growth (Geraerts and Joosse 1984) and adult life

2 189 span (Calow 1978). To a larval trematode, host reproduction diverts resources away from its own development, and potentially limits the size and stability of the parasite's habitat (Bayne and Loker 1987); thus, it is not surprising that trematodes castrate their snail hosts. The intrinsic interdependence existing among the growth, fecundity, and survival components of a life history strategy suggests that parasitic castration could a ect the resulting growth and survival of individual hosts. Increased host size, termed ``gigantism,'' commonly coincides with trematode parasitism (Keymer and Read 1991). Parasitic castration typically begins 2±3 weeks after infection. The cause of this reduction in fecundity may depend upon the type of intramolluscan larval stages that prevail in the trematode life cycle. There are two types of intramolluscan stages of trematodes, sporocysts and rediae, both of which function to produce additional life cycle stages asexually within the host mollusc. Sporocysts lack ambulatory musculature and absorb host nutrients via their tegument. Rediae, with their muscular pharynx and primitive gut, actively consume and digest host tissues while moving throughout infected hosts. Most detailed studies of life history consequences for infected snails have focused on trematodes that possess sporocysts as the nal intramolluscan stage (Pan 1965; McClelland and Bourns 1969; Minchella and LoVerde 1981; Schrag and Rollinson 1994); therefore, relatively little is known about host life history traits associated with trematodes using redial larvae. This study utilizes L. elodes snails and Echinostoma revolutum, a ubiquitous old world trematode that develops rediae stages in pulmonate snails belonging to the genus Lymnaea (Kanev 1994). We address three questions. (1) Do natural populations of echinostome-infected L. elodes show evidence of gigantism? (2) In controlled laboratory infections, does E. revolutum impact the growth, fecundity and/or survival of L. elodes snails? (3) If parasitic infection alters host life history traits, what is the temporal pattern of this change; furthermore, can life history trade-o s explain this outcome? Materials and methods Snail and trematode descriptions The snail L. elodes ( ˆ L. palustris ˆ Stagnicola palustris ˆ S. elodes) is a common, freshwater pulmonate snail that inhabits a variety of ephemeral and permanent lentic systems. L. elodes is typically described as an annual species with a semelparous pattern of reproduction (Eisenberg 1970; Brown et al. 1985); however, the life history traits of this species vary with population density, nutrient availability, and habitat permanence (Eisenberg 1966; Hunter 1975; Brown et al. 1985). Voltinism, growth rate, and size at maturity increase in permanent productive habitats, whereas, life span and reproductive rate are greatest in vernal, densely populated habitats. Because L. elodes is simultaneously hermaphroditic, it can reproduce through either self-insemination or outcrossing. The trematode, E. revolutum, has a complex three-host life cycle and adults use avian species, primarily waterfowl, as de nitive hosts (Kanev 1994). E. revolutum adults, occupying the posterior digestive tract of infected birds, produce eggs that give rise to a free-living stage, called a miracidium, that infects L. elodes snails in northern Indiana, USA (Sorensen et al. 1997). Miracidia of E. revolutum that successfully infect L. elodes produce three distinct asexual stages of the parasite, a mother sporocyst and two subsequent redial stages. The second redial generation produces an infective, free-living cercarial stage that exits the host snail and forms metacercariae in another aquatic host, typically another snail (Beaver 1937). Metacercariae develop into adults when ingested by a suitable vertebrate host. Field survey studies Between 12 July and 9 August 1995, we collected L. elodes from six wetlands in northern Indiana to assess the prevalence of echinostome parasitism within various snail populations. Five of these sites ± pond A, pond B, pond 39, Merry Lea Pond, and Garwood Pond ± are located within Whitley County, Ind., whereas Shock Lake is in Kosciusko County, Ind. These six wetlands present a range of size, permanence, predominant vegetation, and nutrient productivity characteristics (Table 1). Two of these sites (pond A and pond B) have been previously studied (Brown et al. 1985; Minchella et al. 1985). L. elodes tend to reside on vegetation and detritus near the water surface, so we surveyed each study population using a single, haphazard collection of adult-sized L. elodes found on submerged and emergent subtrates. The number of snails collected at each site varied between 24 and 482 depending upon their abundance at the time of collection. All collected snails were isolated in small Syracuse dishes for 12 h under 700 ft-c of illumination to check for emerging echinostome cercariae (see Minchella et al. 1985). We also measured snail size, from the tip of the spire to the base of the body whorl to the nearest 0.01 mm with a digital caliper because shell length was shown to be a reliable indicator of snail condition in previous studies (Hunter 1975; Rollo and Hawryluk 1988). Snail life history traits following experimental infections E. revolutum cercariae were obtained from the infected snails collected during the survey portion of this study in order to experimentally infect snails in the laboratory. Parasites were maintained in the laboratory by passage through chickens (Gallus gallus dom.) and their natural host snails (Sorensen et al. 1997). We collected 500 L. elodes snails, ranging in size from 12 to 23 mm, on 17 June 1996 from Shock Lake. These snails were individually isolated under illumination twice (18±20 June and 2±3 Table 1 Environmental characteristics of the six sites where echinostome-infected and uninfected Lymnaea elodes snails were collected during 1995 in northern Indiana, USA Pond Hectares Permanence Vegetation type Productivity Pond B <0.1 Temporary Grasses Low Merry Lea <0.1 Semipermanent Deciduous Intermediate Pond Temporary Deciduous Intermediate Garwood 0.4 Semipermanent Deciduous Intermediate Pond A 1.6 Temporary Deciduous Intermediate Shock Lake 2.0 Permanent Grasses High

3 190 July) to identify snails that were releasing trematode cercariae. Infected snails were removed from the study population and uninfected snails were housed in several aquaria until 3 July The overall prevalence of patent trematode infections among these snails was 5%; three trematode species were detected: E. revolutum (48%), an unidenti ed trematode possessing amphistome cercariae (32%), and an unidenti ed spirorchid species (20%). No multiple infections were observed among these snails. On 3 July 1996, 118 of the uninfected Shock Lake L. elodes, ranging in size from to mm, were each isolated in 400- ml glass jars to determine their weekly egg production. All snails were fed green leaf lettuce ad libitum throughout the experiment and their water was replaced every 7 days with aged well water. On 11 July 1996, snails were separated into two treatment groups: control snails (n ˆ 26, x size SE ˆ 19:47 0:87, x fecundity SE ˆ 92:38 59:74), and E. revolutum-exposed snails (n ˆ 92, x size SE ˆ 19:47 0:07, x fecundity SE ˆ 96:18 40:70). All snails were placed individually into 10-ml vials containing 9 ml of aged well water. For the snails in the exposed treatments, we added ten miracidia to the vial. Control snails received no additional treatment beyond the isolation into vials. All snails were left in vials for 4 h and then returned to their 400-ml jar. We recorded the shell length, fecundity, and mortality of all snails for 15 weeks. Snail size was measured once each week. Egg masses were removed from the snail jars twice a week throughout the study period and the number of eggs in each mass was recorded. Because growth of L. elodes decreases with age and size (Brown et al. 1985; R.E. Sorensen, unpublished observations), weekly sizedependent growth increments were calculated for all snails according to: growth (t, t)1) ˆ (size t ) size t)1 )/size t)1.whenever a snail died, it was necropsied to determine its infection status. We estimated the quantity of rediae and metacercariae present and assessed the overall condition of the ovotestis/digestive gland (ODG) region, the site where E. revolutum larvae develop. We isolated all snails beginning at 3 weeks post exposure (PE) and twice each week thereafter to ensure that no prepatently infected snails were included in the study populations. The 49 days between the collection date of these snails and the 3-week PE shedding date provided ample time to ensure that the remaining snails contained no prepatant infections, because the trematodes present at this site all reach the cercariae stage in less than 7 weeks. In subsequent snail isolations, when cercariae were observed, a qualitative estimate was made of the number of cercariae present. Once the infection status of snails was determined based on release of cercariae or the presence of rediae, they were classi ed as either infected (INF), exposed but uninfected (EBU), or control (CTL); prior to this, they were classi ed as exposed (EXP) or CTL. Statistical analyses We used parametric tests provided the assumptions of normality and homoscedasticity were met, otherwise we employed non-parametric procedures. An a of 0.05 was used to evaluate statistical signi cance, except when sequential tests were performed, in which case a ``tablewide'' a-value was calculated (Rice 1989). Because previous studies have shown that life history characteristics vary from pond to pond for L. elodes and because size heteroscedasticity existed at several of the sites across the infected and uninfected snails, we utilized the appropriate t-tests (either assuming equal variances or unequal variances) to assess di erences in mean size for eld-collected snails in the two infection classes for each of the six collection sites rather than a two-way ANOVA. However, factorial ANOVA tests were used to test hypotheses involving di erences at a single point in time among the experimentally infected snails. When multiple pairwise comparisons were involved in the ANOVA, the Bonferroni/Dunn post hoc procedure was used to evaluate di erences between means. We employed repeated-measures pro le analysis, MANOVAR (Potvin et al. 1990), when sequential data were collected on the same individuals over time (von Ende 1993). We employed non-parametric tests (Mann-Whitney U and Kruskal-Wallis) on snail size data after 3 weeks PE and on fecundity data after 5 weeks PE because those data violated parametric assumptions after that time. When Kruskal-Wallis tests showed signi cant di erences, we conducted multiple pairwise nonparametric comparisions (Zar 1984). We tested for associations between exposure group and infection status or mortality status with G-tests. Results Field survey studies Based on the release of echinostome cercariae during isolation, 267 (32.4%) of the snails collected from the six populations possessed patent echinostome infections. The prevalence of echinostome infections ranged from 57.6% at Merry Lea Pond to 17.1% at Shock Lake (Table 2). The prevalence of echinostome infections was related to the predominant vegetation type with a greater proportion of infected snails being found in ponds with deciduous vegetation dispersed throughout the littoral zones when compared to sites inhabited primarily by grass species, such as Typhus (G ˆ 12.86, P < 0.001). The mean size of echinostome-infected snails was signi cantly larger than that of uninfected L. elodes at ve of the six wetlands (Table 2). Infected and uninfected groups of snails from Shock Lake had equal mean sizes, and these snails were more than 5 mm smaller than the average size of snails at the other study sites (Table 2). When we compared the size distributions for infected and uninfected snails from Garwood Pond and Shock Lake, we found that all the infected snails at Garwood Pond were larger than the mean size for uninfected snails at that site, whereas infected snails from Shock Lake were distributed across the range of sizes for uninfected snails (Fig. 1). Table 2 Mean size of echinostome-infected and uninfected L. elodes snails collected during 1995 at various sites in northern Indiana, USA (values in parentheses are sample sizes) Pond Date Mean snail size (mm) t df P Infected Uninfected Pond A July (170) (312) <0.001 Pond B July (8) (30) Merry Lea July (19) (14) Pond 39 July (50) (128) <0.001 Garwood August (8) (16) Shock Lake August (12) (58)

4 191 Snail life history traits following experimental infections Snail infection and mortality We rst detected E. revolutum larvae at 29 days PE when two necropsied L. elodes snails contained rediae. Echinostome cercariae were rst observed during snail isolations at 32 days PE. Based on the eventual release of cercariae or the presence of rediae, it was determined that 56.5% (52) of the EXP snails became infected with E. revolutum. There was no relationship between initial snail size and infection status. E. revolutum miracidia increased mortality of EXP L. elodes relative to control snails during the early stages of infection (G ˆ 6.2, P ˆ 0.013). Between weeks 1 and 4 PE, 14.1% of the EXP and 7.7% of the CTL snails died. All of the CTL snails that died did so during the rst week PE, whereas the EXP snails died throughout the 4-week period PE. E. revolutum infection also decreased host survival after 4 weeks PE (Fig. 2). INF snails showed signi cantly higher mortality than uninfected snails (both EBU and CTL combined) at 6, 7, and 8 weeks PE (G ˆ 5.3, P ˆ 0.02; G ˆ 14.3, P < 0.001; G ˆ 18.4, P < 0.001, respectively). The EBU L. elodes survived at rates comparable to CTL snails (Fig. 2). No INF snails survived beyond 10 weeks (Fig. 2); furthermore, the proportion of INF snails that survived through the 8th week PE was signi cantly smaller than the proportion of surviving EBU and CTL snails (G ˆ 16.7, P < 0.001). Necropsies of infected snails and cercariae release patterns showed common trends over time. After 29 days, we found few rediae (<30) in the ODG and few cercariae were released over a 2-h period (<50). During later weeks, the number of rediae and cercariae that were observed increased by one to two orders of magnitude. The growth of the rediae population had a strong negative in uence on the ODG as rediae apparently consumed ODG tissue, causing its size and structural integrity to decrease over time. The ODG of most hosts that died after 7 weeks PE was virtually destroyed. With the increasing number of cercariae being produced by these rediae, there was a corresponding increase in the number of metacercariae present in their hosts. Because snails were individually isolated, cercariae released from the host reentered the snail's heart and kidney tissues producing a superinfection. Snail fecundity An E. revolutum infection had a strong e ect on L. elodes fecundity; egg production decreased sharply beginning at 3 weeks PE (Fig. 3). INF snails had signi cantly lower weekly egg output compared to the two uninfected snail classes (EBU and CTL) at 5 weeks PE (F 2,92 ˆ 23.71, P < 0.001; infected x ˆ 31:97, uninfected x ˆ 234:25, control x ˆ 146:59). Mean fecundity values for EBU and CTL snails did not di er during that time period. The reduction in egg production among parasitized snails continued, and fewer hosts laid eggs over time until eventually all egg production ceased after 7 weeks PE (Fig. 3). Among snails in the two uninfected classes (EBU and CTL), fecundity was variable beyond 3 weeks PE, but there was a general negative trend as time increased Fig. 1 Size distribution of echinostome-infected and uninfected Lymnaea elodes snails collected 10 August 1995 at Garwood Pond (A) and on 9 August 1995 at Shock Lake marsh (B). The mean size of uninfected snails at Garwood Pond and Shock Lake is mm and mm, respectively Fig. 2 Percent survival of L. elodes snails at 4±15 weeks following exposure to Echinostoma revolutum miracidia. Infection status was based on rediae or cercariae presence after autopsies of dead snails or isolation of live snails under illumination at 4±7 weeks post exposure. Data points do not re ect deaths that occured 0±4 weeks post exposure

5 192 Fig. 3 Average fecundity of L. elodes snails in three treatment groups following exposure to E. revolutum miracidia. No infected snails survived beyond week 10. Error bars display the standard error of the estimate of mean weekly fecundity for each group Fig. 4 Average size of L. elodes snails in three experimental groups. No infected snails survived beyond week 10. Error bars display the standard error of the estimated mean size for each group (Fig. 3). Most of these snails continued to lay eggs throughout the experiment. For instance, 13 of the 18 EBU and CTL snails that survived until week 15 PE were still producing eggs at that time. Snail growth The average shell length of snails in the three groups (INF, EBU, and CTL) increased over time through 3 weeks PE as shown by the Wilks' lambda test associated with a MANOVAR phase analysis of a time e ect (F 3,99 ˆ 86.70, P < ). Furthermore, di erences in growth rate were not observed between the snails in the three experimental groups because no time experimental group e ect was noted during this 3-week period (F 6,198 ˆ 0.31, P ˆ 0.93). After 3 weeks PE, however, the mean size for INF snails increased rapidly compared to either EBU and CTL snails (Fig. 4). At 6 weeks PE, a signi cant size di erence between INF, EBU, and CTL snails was noted (H ˆ 8.9, P < 0.001). Multiple pairwise comparisons showed that INF snails were signi cantly larger than either EBU (Q ˆ 2.7, P < 0.05) and CTL snails (Q ˆ 3.2, P < 0.005). This di erence in mean shell length among the three groups remained throughout the 10 weeks PE that INF snails survived (Fig. 4). INF snails were nearly 1 mm larger, on average, than snails in the EBU and CTL classes at 5 weeks, and >2 mm larger at 10 weeks PE. Altered growth patterns were also observed among EBU snails beyond 5 weeks PE. Mean shell length of EBU snails was always greater than that of CTL snails (Fig. 4). At week 9 PE, EBU snails were signi cantly larger than CTL snails (time treatment, F 1,9 ˆ 2.674, P ˆ 0.005). After 9 weeks PE, the EBU snails were >1 mm larger than CTL snails, on average, and this di erence increased slightly by week 15 (Fig. 4). Either divergent shell growth rates or size-speci c mortality could have yielded mean size di erences among snails in the treatment groups. To test for growth rate di erences across treatments, the size-dependent weekly growth increment was calculated for weeks 3 through 8 PE. We found signi cant di erences in weekly growth rate for all weeks except week 5 PE (for weeks 3, 4, 6±8, H > 9.0, F < 0.04; for week 5, H ˆ 4.7, P ˆ 0.09). Subsequent multiple comparisons showed that INF snails increased their shell length to a greater extent than either the EBU or CTL snails during weeks 3 and 4 PE (Q ˆ 3.3, P < and Q ˆ 3.3, P < 0.05, respectively). Beyond this time, EBU snails added shell material more quickly than the INF snails (Q ˆ 3.3, P < 0.005; Q ˆ 2.9, P < 0.05; Q ˆ 3.5, P < 0.005, for weeks 6, 7, and 8, respectively); thus, the increase in mean size of INF snails relative to EBU or CTL snails beyond week 5 PE was caused by size-speci c mortality, not by an increased growth rate. Larger INF snails survived longer than smaller INF snails in weeks 5 and 7 PE (U ˆ 25.0, P ˆ and U ˆ 132.0, P ˆ 0.015, respectively). This e ect was not found in the EBU and CTL snails. Discussion Parasitism by E. revolutum has a signi cant in uence on L. elodes growth, fecundity, and survival. We found that snails exposed to E. revolutum su ered increased mortality compared to control snails. Furthermore, infected

6 193 snails showed size-speci c mortality favoring larger individuals, grew faster early in the infection, and displayed reduced fecundity relative to uninfected snails. Snail infection and mortality Both exposure to and infection by E. revolutum markedly reduced L. elodes survival as also demonstrated in several other freshwater trematode-snail systems (Anderson and May 1979; Loker 1979; Kuris 1980; Minchella and LoVerde, 1981). We detected two phases of host mortality: during the prepatent stage (weeks 0±4), and several weeks postpatency (Fig. 2). Host mortality during the prepatent stage may have resulted from increased energetic demands of host defense, or from starvation, because tissue degradation was not obvious. In contrast, mortality of older snails involved tissue degradation. Superinfection with metacercariae is another potential cause of reduced survival in host snails during the late stages of infection; however, little is known about the relationship between metacercariae burden and mortality (Beaver 1937). The function of host immune responses was not investigated here, but initiation of these responses must be energetically demanding and the cost may depend upon the number of miracidia or metacercariae that enter the snail (Bayne and Loker 1987). Snail fecundity Trematode parasitism typically decreases the fecundity of hosts during the patent stages of an infection. Our results demonstrate that ``parasitic castration'' occurs in this parasite-host system (Fig. 3). The precise cause of this reduction is unknown, but it seems unlikely that physical destruction of host tissues is the sole mechanism (Wilson and Denison 1980), since few (<30) rediae were found in necropsied snails that died early in patency. Furthermore, the digestive gland and ovotestis of snails displayed no evidence of damage until later in the patency (5+ weeks). These results parallel those of Cheng et al. (1973), who concluded that chemical factors rather than physical damage were the cause of ``parasitic castration'' in Ilyanassa obsoleta snails infected with the sporocyst trematode, Zoogonus rubellus. During the early stages of an infection (0±4 weeks), before reproductive tissues become damaged, it is possible that snail fecundity is reduced in this host-parasite system simply because of a reduction in the amount of energy accessible to the snail given the energetic demands associated with parasitism. Rollo and Hawryluk (1988) showed that uninfected L. elodes snails fed a lowquality diet spent more time foraging and reduced the amount of energy allocated to reproduction relative to snails on a high-quality control diet. It has been shown that schistosome infections reduce the glucose and amino acid levels of host snails in a manner similar to starvation (Christie et al. 1974; Stanislawski et al. 1979; Trede and Becker 1982); therefore, if echinostome infections deplete host resources, the reduction in nutrient availability may provide the impetus for altered fecundity patterns among infected L. elodes. Minchella and LoVerde (1981) documented a process whereby schistosome-infected snails compensate for future reproductive losses due to parasitism by increasing their fecundity during the prepatent period of an infection. More recently, this ``fecundity compensation'' mechanism has been reported in two other schistosome systems involving distantly related planorbid and lymnaeid snail hosts (Thornhill et al. 1986; Schallig et al. 1991). We found no evidence of ``fecundity compensation'' among E. revolutum-infected L. elodes snails; the mean number of eggs laid per week by infected snails did not exceed that of uninfected or control snails during the prepatent period. Although few detailed studies of other trematode systems are available for comparison, it appears that ``fecundity compensation'' may be limited to trematode systems that utilize sporocysts as the predominant intramolluscan stage. Thus, schistosomes may mimic the conditions of Rollo and Hawryluk's (1988) intermediate-quality diet which yielded a temporary increase in snail reproduction relative to well-fed individuals. Energy usage by E. revolutum rediae may not leave su cient host resources to elicit a ``fecundity compensation'' response. Therefore, energy depletion/allocation processes could explain the apparent incongruity between precastration fecundity among schistosome- and echinostome-infected snails. Snail growth Brown et al. (1988) demonstrated a positive relationship between the prevalence of trematode infections and snail size among three populations of L. elodes in Indiana. The same relationship holds at ve of the six sites we surveyed, as the shell length of echinostome-infected L. elodes was signi cantly greater than that of uninfected L. elodes (Table 2, Fig. 1). Baudoin (1975) presented six alternative hypotheses to explain the positive correlation between host size and the prevalence of infection. These hypotheses involve three basic mechanisms: increased host growth rates, di erential host mortality rates, and size-speci c infection preferences of parasites. Our data suggest that no single hypothesis accounts for gigantism in this system. The signi cant increase in adult host size that occurred as a result of the experimental infections (Fig. 4) supports the growth rate hypothesis during early stages of a trematode infection and the size-speci c mortality hypothesis later in the infection. It remains to be determined whether or not age-speci c infection rates are involved in this system, as shown by Curtis (1995) with trematode-infected I. obsoleta. Sousa (1983) predicted that among short-lived, semelparous mollusc species, gigantism will occur following trematode parasitism because excess host energy

7 194 reserves are made available by ``parasitic castration.'' This prediction requires that the energetic demands of parasitism are less than the costs of reproduction in the absence of parasitism (Bourns 1974). Data presented here support Sousa (1983) and Bourns (1974) by showing that L. elodes growth rates increase concurrently with depressed fecundity (Figs. 3, 4). This outcome suggests that host growth increases early in an infection because larval parasite development extracts su cient energy to reduce host fecundity, but not so much that growth is prevented. In that sense, increased growth following an E. revolutum infection can be considered a manifestation of the trade-o between L. elodes fecundity and growth. Three evolutionary hypotheses have been proposed as ultimate causes of gigantism among sterilized adult hosts: a parasite advantage, a host advantage, or a consequence of the life history trade-o between host growth and fecundity (Dawkins 1982; Minchella 1985; Minchella et al. 1985). In many cases, these alternative hypotheses cannot be clearly distinguished by considering only the proximate outcome of parasite-host interactions due to the inextricable link between host fecundity and growth. Also, it may be di cult to unambiguously assign tness costs and bene ts experienced by hosts or parasites following a change in the host's life history. We feel that the host advantage hypothesis is supported by the elevated growth of our exposed but uninfected snails relative to control snails because exposure to miracidia in the absence of an infection elicits enhanced growth rates among exposed snails. Growth is presumably the predominant energy demand for juvenile gastropods; as a result, infected immature snails should grow more slowly than uninfected conspeci cs (Sousa 1983). This prediction has been supported in several studies of infected lymnaeid snails (Zischke 1967; Loker 1979). For instance, Zischke (1967) demonstrated that uninfected, juvenile L. elodes grow more rapidly than individuals harboring echinostome infections. On the other hand, our results indicate that an increase in shell length occurs in adult snails following infection. Therefore, within natural L. elodes populations, it is likely that infected, immature L. elodes grow slower, on average, than uninfected conspeci cs, while parasitized adult snails grow faster than uninfected snails. It seems incongruous that gigantism was not observed among naturally infected Shock Lake snails, but it was observed among the experimentally infected snails which originated from Shock Lake. The age-speci c parasite in uence on L. elodes growth rates helps explain the lack of gigantism among infected snails collected during the 1995 survey at Shock Lake. This site is a productive, permanent wetland that provides L. elodes with the conditions to allow a cohort to reach reproductive maturity during their rst year; thus, when sampled, this population contained both mature (15 mm) and immature members of the 1995 cohort. It could be that snails infected prior to maturity su ered reduced growth rates, whereas snails infected as adults experienced increased growth rates. Thus, under these environmental conditions, a bimodal distribution of host size would be expected, and the mean values for these two size distributions would bracket the mean size of uninfected snails. The net e ect of these two agespeci c processes would be a similarity in mean size for infected and uninfected snails, as we found. This process requires that the infected L. elodes were parasitized during the summer we collected them. This appears likely based on the reduced survival shown by our experimentally infected snails and by prevalence surveys for an ecologically similar freshwater pulmonate, Helisoma trivolvis, that show echinostome parasitism peaking in late summer and waning throughout the winter (Schmidt and Fried 1997). Evidence of age-speci c growth rates was not found at our other collection sites because their ephemerality and limited productivity prevents new cohorts from reaching adult size by the time the survey collections were made. In other words, the snails collected at the sites where gigantism was detected were members of the previous year's cohort rather than snails produced during Presumably, these hosts became infected following maturity which led to increased growth following ``parasitic castration'' and the e ects of size-speci c mortality. This outcome demonstrates a potential bias that can result when considering the impact of trematode infections on host growth in natural populations (Fernandez and Esch 1991). In nature, the variability of host growth rates may increase markedly following parasitism if the growth rates of mature and immature individuals are a ected di erently. Acknowledgements This work was supported by the Purdue Research Foundation, Sigma Xi, and the Alton A. Lindsey Graduate Fellowship. We thank P.J.C. Sorensen for invaluable assistance with the experimental work. Furthermore, we appreciate the thoughtful advice o ered by Peter Fauth, Richard Howard, Je rey Lucas, and two anonymous reviewers who greatly improved the clarity of this manuscript. References Anderson RM, May RM (1979) Prevalence of schistosome infections within molluscan populations: observed patterns and theoretical predictions. Parasitology 79: 63±94 Baudoin M (1975) Host castration as a parasitic strategy. Evolution 29: 335±352 Bayne CJ, Loker ES (1987) Survival within the snail host. In: Rollinson D, Simpson AJG (eds) The biology of schistosomes: from genes to latrines. Academic Press, San Diego, pp 321±346 Beaver PC (1937) Experimental studies on Echinostoma revolutum (Froelich): a uke from birds and mammals. Ill Biol Monogr 15: 1±96 Bourns TKR (1974) Carbohydrate and protein in Lymnaea stagnalis eggs and Trichobilharzia ocellata cercariae. J Parasitol 60: 1046±1047 Brown KM (1982) Resource overlap and competition in pond snails: an experimental analysis. Ecology 63: 412±422 Brown KM, DeVries DR, Leathers BK (1985) Causes of life history variation in the freshwater snail Lymnaea elodes. Malacologia 26: 191±200

8 195 Brown KM, Leathers BK, Minchella DJ (1988) Trematode prevalence and the population dynamics of freshwater pond snails. Am Midl Nat 120: 289±301 Calow P (1978) The evolution of life-cycle strategies in freshwater gastropods. Malacologia 17: 351±364 Cheng TC, Sullivan JT, Harris KR (1973) Parasitic castration of the marine prosobranch gastropod Nassarius obsoletus by sporocysts of Zoogonus rubellus (Trematoda): histopathology. J Invert Pathol 21: 183±190 Christie JD, Foster WB, Stauber LA (1974) The e ect of parasitism and starvation on carbohydrate reserves of Biomphalaria glabrata. J Invert Pathol 23: 55±62 Crowl TA, Covich AP (1990) Predator-induced life-history shifts in a freshwater snail. Science 247: 949±951 Curtis LA (1995) Growth, trematode parasitism, and longevity of a long-lived marine gastropod (Ilyanassa obsoleta). J Mar Biol Assoc UK 75: 913±925 Dawkins R (1982) The extended phenotype. Freeman, Oxford Eisenberg RM (1966) The regulation of density in a natural population of the pond snail Lymnaea elodes. Ecology 47: 889±906 Eisenberg RM (1970) The role of food in the regulation of the pond snail, Lymnaea elodes. Ecology 51: 680±684 Ende CN von (1993) Repeated-measures analysis: growth and other time-dependent measures. In: Scheiner SM, Gurevitch J (eds) Design and analysis of ecological experiments. Chapman and Hall, New York, pp 113±137 Esch GW, Fernandez JC (1994) Snail-trematode interactions and parasite community dynamics in aquatic systems: a review. Am Midl Nat 131: 209±237 Fernandez J, Esch GW (1991) E ect of parasitism on the growth rate of the pulmonate snail Helisoma anceps. J Parasitol 77: 937±944 Geraerts WPM, Joosse J (1984) Freshwater snails (Basommatophora). In: Salueddin ASM, Wilbur KM (eds), The mollusca, vol 7. Academic Press, New York, pp 141±207 Hunter RD (1975) Growth, fecundity, and bioenergetics in three populations of Lymnaea palustris in upstate New York. Ecology 56: 50±63 Kanev I (1994) Life-cycle, delimitation and redescription of Echinostoma revolutum (Froelich, 1802) (Trematoda: Echinostomatidae). Syst Parasitol 28: 124±144 Keas BE, Esch GE (1997) The e ect of diet and repoductive maturity on the growth and reproduction of Helisoma anceps (Pulmonata) infected by Halipegus occidualis (Trematoda). J Parasitol 83: 96±104 Keymer AE, Read AF (1991) Behavioral ecology: the impact of parasitism. In: Toft CA, Aeschlimann A, Bolis L (eds) Parasitehost associations: coexistence or con ict? Oxford University Press, New York, pp 37±61 Kuris AM (1980) E ect of exposure to Echinostoma liei on growth and survival of young Biomphalaria glabrata snails. Int J Parasitol 10: 303±308 Loker ES (1979) E ects of Schisomatium douthitti infection on the growth, survival, and reproduction of Lymnaea catascopium. J Invert Pathol 34: 138±144 Malek EA, Cheng TC (1974) Medical and economic malacology. Academic Press, New York. pp 204±241 McClelland G, Bourns TKR (1969) E ects of Trichobilharzia ocellata on growth, reproduction, and survival of Lymnaea stagnalis. Exp Parasitol 24: 137±146 Michalakis Y, Hochberg ME (1994) Parasitic e ects on host lifehistory traits: a review of recent studies. Parasite 1: 291±294 Minchella DJ (1985) Host life history variation in response to parasitism. Parasitology 90: 205±216 Minchella DJ, LoVerde PT (1981) A cost of increased early reproductive e ort in the snail Biomphalaria glabrata. Am Nat 118: 876±881 Minchella DJ, LoVerde PT (1983) Laboratory comparison of the relative success of Biomphalaria glabrata stocks which are susceptible and insusceptible to infection with Schistosoma mansoni. Parasitology 86: 335±344 Minchella DJ, Leathers BK, Brown KM, McNair JN (1985) Host and parasite counteradaptations: an example from a freshwater snail. Am Nat 126: 843±854 Pan CT (1965) Studies on the host-parasite relationship between Schistosoma mansoni and the snail Australorbis glabratus. Am J Trop Med Hyg 14: 931±976 Potvin C, Lechowicz MJ, Tardif S (1990) The statistical analysis of ecophysiological response curves obtained from experiments involving repeated measures. Ecology 71: 1389±1400 Price P (1980) Evolutionary biology of parasites. Princeton University Press, Princeton Rice WR (1989) Analyzing tables of statistical tests. Evolution 43: 223±225 Rollo CD, Hawryluk MD (1988) Compensatory scope and resource allocation in two species of aquatic snails. Ecology 69: 146±156 Schallig HDFH, Sassen MJM, Hordijk PL, De Jong-Brink M (1991) Trichobilharzia ocellata: in uence of infection on the fecundity of its intermediate snail host Lymnaea stagnalis and cercarial induction of the release of schistosomin, a snail neuropeptide antagonizing female gonadotropic hormones. Parasitology 102: 85±91 Schmidt KA, Fried B (1997) Prevalence of larval trematodes in Helisoma trivolvis (Gastropoda) from a farm pond in Northampton County, Pennsylvania with special emphasis on Echinostoma trivolvis (Trematoda) cercariae. J Helminthol Soc Wash 64: 157±159 Schrag SJ, Rollinson D (1994) E ects of Schistosoma haemotobium infection on reproductive success and male outcrossing ability in the simultaneous hermaphrodite, Bulinus truncatus (Gastropoda: Planorbidae). Parasitology 108: 27±34 Sorensen RE, Kanev I, Fried B, Minchella DJ (1997) The occurrence and identi cation of Echinostoma revolutum from North American Lymnaea elodes snails. J Parasitol 83: 169±170 Sousa WP (1983) Host life history and the e ect of parasitic castration on growth: a eld study of Cerithidea californica Haldeman (Gastropoda: Prosobranchia) and its trematode parasites. J Exp Mar Biol 73: 273±296 Stanislawski E, Becker W, Muller G (1979) Alterations of the free amino acid content in the hemolymph of Biomphalaria glabrata (Pulmonata) in starvation and after infection with Schistosoma mansoni (Trematoda). Comp Biochem Physiol 63B: 477±482 Stearns SC (1976) Life history tactics: a review of the ideas. Q Rev Biol 51: 3±47 Stearns SC (1992) The evolution of life histories. Oxford University Press, New York Thompson SN (1997) Physiology and biochemistry of snailtrematode relationships. In: Fried B, Graczyk TK (eds) Advances in trematode biology. CRC Press, Boca Raton, Fla., pp 149±196 Thornhill JA, Jones JT, Kusel JR (1986) Increased oviposition and growth in immature Biomphalaria glabrata after exposure to Schistosoma mansoni. Parasitology 93: 443±450 Toft CA (1991) An ecological perspective: population and community perspectives. In: Toft CA, Aeschlimann A, Bolis L (eds) Parasite-host associations: coexistence or con ict? Oxford University Press, New York, pp 319±343 Trede G, Becker W (1982) E ects of starvation and infection with Schistosoma mansoni on the relaease rate of free amino acids (FAA) of Biomphalaria glabrata. Comp Biochem Physiol 73B: 405±409 Wilson RA, Denison J (1980) The parasitic castration and gigantism of Lymnaea truncatula infected with the larval stages of Fasciola hepatica. Z Parasitenkd 61: 109±119 Wright CA (1971) Flukes and snails. Allen and Unwin, London Zar JH (1984) Biostatistical analysis. Prentice Hall, Englewood Cli s, NJ Zischke JA (1967) Redial populations of Echinostoma revolutum developing in snails of di erent sizes. J Parasitol 53: 1200± 1204