Introduction
Host quality for parasitoid growth and development has often been assumed to scale with host size as host body size determines the maximum amount of food resources available for the developing parasitoid (Charnov et al., Reference Charnov, Los-den-Hartogh, Jones and van den Assem1981; King, Reference King, Wrensch and Ebert1993; Godfray, Reference Godfray1994). Most studies supporting this trend, however, are based on idiobiont parasitoids that attack non-feeding host stages (eggs or pupae) or paralyzed hosts (Corrigan & Lashomb, Reference Corrigan and Lashomb1990; Otto & Mackauer, Reference Otto and Mackauer1998). In contrast, koinobiont parasitoids attack hosts that continue to feed, grow and develop after it has been parasitized, with parasitism often occurring at an early stage of host development. Thus, the future resources for the developing parasitoid larva can vary depending on the host's age or stage of development, rather than on its size at the time of parasitoid oviposition (Mackauer et al., Reference Mackauer, Sequeira, Otte, Detter, Bauer and Volkl1997; Harvey, Reference Harvey2000). Host quality for koinobiont parasitoids is, thus, crucially dependent on the rate of host growth after parasitism and on the final size of the host when it is destroyed by the parasitoid. When attacking nutritionally suboptimal hosts, such as early instars, the host may grow too slowly for the parasitoid to maximize body size and minimize developmental duration. Consequently, the optimal phenotype will be based on a trade-off between these fitness parameters (Harvey & Strand, Reference Harvey and Strand2002). Strand (Reference Strand, Hochberg and Ives2000) proposed the ‘dome-shaped’ model, suggesting that the fitness of many koinobionts are likely to be a function of the host size (and age) at parasitism and the offspring fitness initially increases with host size, but declines rapidly as the host resource is depleted because of consumption of the growing parasitoid larva.
The present study examined the growth and developmental interactions of the solitary larval endoparasitoid Meteorus pulchricornis (Wesmael) [=japonicus Ashmead] (Hymenoptera: Braconidae) and its host, the cotton bollworm Helicoverpa armigera (Hübner) (Lepidoptera: Noctuidae). Meteorus pulchricornis is reported to also attack most larval instars of the gypsy moth (Lymantria dispar (L.) (Feuster et al., Reference Fuester, Taylor, Peng and Swan1993), and the beet armyworm Spodoptera exigua (Hübner) (Liu & Li, Reference Liu and Li2006). Little is known about the developmental interactions between this parasitoid and the cotton bollworm, except for a brief description of the biology of the species in southwestern China (Li, Reference Li1984). Our study serves two purposes: (i) to test the model proposed by Strand (Reference Strand, Hochberg and Ives2000) using this new parasitoid-host system and (ii) to gain a better understanding of this less-known model system itself.
Materials and methods
Culture of the parasitoid and host
Wild stock of M. pulchricornis was originally collected in August 2002 as an asexual (thelytokous) population from S. exigua larvae on soybean in a suburb of Nanjing City, Jiangsu Province (32.0°N and 118.7°E). It was reared in an insectary using H. armigera as hosts at 24±1°C and 60±10% RH with a 14:10 h light-dark photocycle. Larvae of H. armigera were collected from a cotton field in Hebei Province in July 1995 and, thereafter, have been maintained on an artificial diet (Shen & Wu, Reference Shen and Wu1995) in the lab. Host larvae were individually reared in 10×1.5 cm glass tubes from hatching until pupation. Pupae were transferred in groups to cages (23×22.5×32 cm high) with 40-mesh nylon organza over wooden frames for adult eclosion and egg laying. We provided gauzes as a substrate for egg deposition and 10% sucrose solution as supplementary food for the moths. The M. pulchricornis population was maintained in clear plastic containers (15 cm diameter×10 cm high) containing ~40 L2 or L3 host larvae; three adult wasps were released into these containers for about eight hours. Wasps used in the experiment were 4–6 days old (our previous observation showed that the wasp attained high daily oviposition at this age) and had not previously been exposed to host larvae. The final instar larva of the parasitoid chews a hole, exits the hosts and spins a distinctive cocoon, which hangs from a line of silk. The host larva dies within a few hours of emergence of the parasitoid.
Experiment
H. armigera larvae undergo six instars (designated L1–L6) before pupating. The mean weight and age of host larvae on the day when each age-cohort was parasitized are given in table 1. A cohort of 20 larvae of each instar within five hours of molting were individually weighed to 0.01 mg on a microbalance (LAC114, Lavrock, Changshu) and then transferred into a clear plastic container (11 cm in diameter and 6 cm in height) with a 3 cm hole in the lid covered by nylon organza for ventilation and another hole on the side (1 cm) for introducing wasps. A fresh cotton leaf was spread on the bottom of the container and with 1% agar below the leaf to keep it fresh. An adult parasitoid of 4–6 days old was released into the container with 20 host larvae and left to attack hosts for eight hours. This time interval was sufficient for maximum parasitism but not superparasitim as demonstrated by an exploratory experiment. A honey solution of 10% was provided for the parasitoid. The host larvae were then reared singly in glass tubes on artificial diet until they pupated if not parasitized, or until parasitoid larvae egressed if parasitized. Twenty replicates (parasitoids) were used for each instar cohort of host larvae. The experiment was conducted at 26±1°C, 60±10% RH under a photoperiod of 14:10 (L:D).
Body size and development time of M. pulchricornis when parasitizing different instars of H. armigara larvae.
Means within column followed by the same letter are not significantly different (P>0.05, Tukey's multiple range test on main effects). Means within row followed by * are significantly different between instars (P<0.05). Values in brackets are the number of parasitoid larvae egressed.
Host larvae reared in glass tubes were monitored twice per day (at 8:00 and 20:00 h). Larval molting time was recorded to determine the development time of each instar in days. Host larvae that died midway between instars were dissected to check if the death was caused by parasitism. Parasitoid larval egression from hosts was recorded, cocoons were individually weighed and adult emergence was recorded.
Parasitoid mortality was measured from larval egression to adult eclosion. The development time of the parasitoid was measured as the number of days from oviposition to adult eclosion. The eclosing parasitoids were killed in 70% alcohol and the length of the right hind tibia was measured under a stereo-microscope as a standard measure of body size for parasitoids (Godfray, Reference Godfray1994).
Statistical analysis
Data for parasitoid mortality were compared using logistic regression to determine if the data are non-linear and the data was also compared using x 2-test to determine if the data varies significantly with host age at parasitism. Because data for L1 may exert a disproportional effect on the overall pattern of mortality, logistic regression was performed on mortality data with data for L1 included and excluded. Parasitism of each instar hosts and mortality of offspring parasitoids were arcsine square root transformed prior to analysis to normalize the data set. Because the experimental design included multiple observations on body sizes of offspring parasitoids for each host instar, the linear regression was first tested for departure from linearity before fitting the linear model (Zar, Reference Zar1996). Development times, cocoon weights and hind tibia length of the parasitoids from host instars were compared with ANOVA and Tukeys multiple comparison tests. Development times between unparasitized and parasitized host larvae were compared with t-tests. Relationships between host age and parasitoid development, cocoon weight and length of hind tibia were analyzed by regression analysis. The significance level was 5% unless noted otherwise.
Results
Parasitism of different larval instars of H. armigera
All six larval instars were susceptible to parasitism by M. pulchricornis, but L2, L3 and L4 larvae were significantly more susceptible than L1, L5 and L6 larvae (ANOVA, F 5, 114=37.22, P<0.001, fig. 1). In single choice tests, where the parasitoid females were offered host larvae of different instar cohorts, the wasp parasitized more than 50% of L2 through L4 larvae, more than 30% of L1 and L5 larvae, and only 1.3% of L6 larvae.
Parasitism of M. pulchricornis on different instars of H. armigera larvae. Error bars represent standard errors of means and different letters indicate significant differences (Tukeys multiple test, P<0.05). Sample sizes are: L1, 20; L2, 20; L3, 20; L4, 20; L5, 20; L6, 20.
Parasitoid mortality, development time, cocoon weight and eclosing adult size in different larval instars of H. armigera
Logistic regression revealed that parasitoid mortality was a non-linear function of host age, either when data from L1 hosts was included (x 21=74.37, P<0.001) or excluded (x 21=36.76, P<0.001) from analysis. Furthermore, parasitoid mortality varied significantly with host instar at oviposition (x 2=28.606, P<0.0001). Parasitoids experienced much higher mortality in L1, L6 and L5 host larva than in L2–L4 at the time of parasitism (fig. 2a).
Developmental characteristics of M. pulchricornis on different instars of H. armigera larvae. (a) Mortality from larval egression to adult eclosion. Sample sizes are: L1, 20; L2, 20; L3, 20; L4, 20; L5, 20; L6, 4. (b) Cocoon weight. Sample sizes are: L1, 78; L2, 90; L3, 87; L4, 90; L5, 76; L6, 3. (c) Hind tibia length of adult offspring. Sample sizes are: L1, 78; L2, 90; L3, 87; L4, 90; L5, 76; L6, 3. (d) Development time in days from oviposition to adult eclosion. Sample sizes are: L1, 78; L2, 90; L3, 87; L4, 90; L5, 76; L6, 3. Points with the different letters indicate significant differences (Tukey's HSD tests, P<0.05). Bars represent standard errors of means.
Both parasitoid cocoon weight and final adult size (length of hind tibia) varied significantly with host instar parasitized (for cocoon weight: F 5, 419=125.08, P<0.001; for adult size: F 5, 91=45.47, P<0.001) (fig. 2b, c). There was a significant positive relationship between adult wasp size and host instar parasitized (y=0.0226x+1.4799, n=100, r=0.8082, P<0.001). Wasps, which emerged from L4, L5 and L6 hosts, were 7% larger than those from L1 hosts.
Development time of M. pulchricornis from oviposition to adult eclosion varied significantly depending on which host larval instar was parasitized (F 5, 368=270.968, P<0.001). Development time of parasitoids from older larvae (L4–L6) was significantly longer than those from younger larvae (L1–L3), whereas parasitoids from L1 took significantly longer time to complete development than those from L2 and L3 (fig. 2d). Mature parasitoid larvae egressed from a range of host instars when development was initiated in L1 and L3–L5 hosts (table 1). Most parasitoid larvae came out from L4 (93.2%) rather than L3 hosts (6.8%) when oviposited into L1 hosts. However, more than half of parasitoid larvae egressed from L4 (60.3%) rather than L5 (39.7%) hosts when oviposited into L3 hosts. Most of the larvae exited from L5 (74.5%) and L4 (21.3%), but a few exited from L6 (4.3%) when L4 host larvae were attacked. Much more parasitoid larvae came out from L6 (71%) than from L5 (29%) when L5 larvae were parasitized. For parasitoids oviposited into the same host instars, no significant difference was observed in the development time from oviposition to larval egression, and nor was any difference found in adult hind tibia length, even if the parasitoids emerged from different host instars. The only exception was that the development time for parasitoids egressed from L3 was significantly longer than those from the L4 when L1 was attacked.
Development of the parasitized host larvae
Development time of parasitized host instars varied depending on the instar attacked and time of parasitoid egression. Developmental period was significantly prolonged in the final instar from which the parasitoid egressed no matter in which of the six instars the host was parasitized (table 2).The development time of host instars next to the final instar was about the same compared to that of the respective healthy counterparts when first two instars (L1 and L2) were parasitized, whereas it was prolonged or shortened when the other instars were attacked (table 2).
Developmental period of host instars of H. armigera after parasitized at different instars by M. pulchricornis.
* Significant differences with unparasitized larvae (P<0.05; t-test); **significant difference with unparasitized larvae (P<0.01; t-test).
Discussion
The results of this study demonstrate that for the parasitoid species M. pulchricornis, instar-specific differences in host quality did not only affect the perceptibility (measured as percent parasitism), but also the fitness returns of M. pulchricornis. Most importantly, older and, hence, larger host larvae were simply not better hosts for the developing parasitoid wasps. Although parasitoid size (measured as cocoon weight and adult hind tibia length) was positively correlated with host instar at parasitism, parasitoids developing in larger hosts (L5 and L6) suffered much higher mortality than conspecifics developing in smaller hosts (L2–L4). Furthermore, egg-to-adult development time in of M. pulchricornis was significantly longer in older host larval stagese (L4–L6) than in the younger host stages. The performance of M. pulchricornis parasitoids measured by the performance of fitness-related traits in M. pulchricornis strongly indicates that the L3 host larva is the most suitable for parasitoid survival, growth and development, followed by both L2 and L4 hosts; whereas L1, L5 and L6 are the least favorable hosts. The results of our study, therefore, supported the dome-shaped model of fitness functions of many koinobionts proposed by Strand (Reference Strand, Hochberg and Ives2000) and refined by Harvey & Strand (Reference Harvey and Strand2002).
The increased mortality from egression to adult eclosion in the parasitoid developing in youngest (L1) and oldest (L5, L6) hosts appears to have resulted from different causes. In L1 hosts, the high mortality may have been due to injuries to the host through insertion and removal of the ovipositor at parasitism, which appears to be the case with Venturia canescens and its host, Plodia interpunctella (Harvey et al., Reference Harvey, Harvey and Thompson1994). High parasitoid mortality in the oldest hosts is probably attributable to the ineffective control over the developmental program of older host larvae by immature parasitoids, as some lethal factors in older host larvae, such as the presence of host toxins and infection, environmental factors and contact of the host with xenobiotic factors are beyond the control of the parasitoid (Vinson & Iwantsch, Reference Vinson and Iwantsch1980).
Strand (Reference Strand, Hochberg and Ives2000) argued that koinobionts have evolved strategies that either make host resources more predictable or reduce the costs of imprecisely allocating offspring to hosts. The results of our study did not provide support for the first mechanism, i.e. to make host resources more predictable, because the parasitoids would oviposit into host larvae of all six instars. Furthermore, M. pulchricornis displayed an oviposition tendency, represented by parasitism, not perfectly consistent with the performance of the offspring. On the other hand, the same parasitism levels amongst L2–L4 hosts, nonetheless, have resulted in differences in fitness-related performance in offspring parasitoids (fig. 2), thus providing support for Strand's (Reference Strand, Hochberg and Ives2000) second mechanism, i.e. to reduce the costs of imprecisely allocating offspring to hosts, and is in accordance with the hypothesis that the maintenance of a flexible host range, where the range of host acceptability is broader than the range of host suitability, should be a more effective long-term strategy in ensuring that the parasitoid would be able to respond effectively to minor phenotypic changes in the host (Mackauer, Reference Mackauer and Lowe1973; Gauld, Reference Gauld1988; Strand, Reference Strand, Hochberg and Ives2000).
Koinobiont parasitoids have evolved a variety of adaptive mechanisms to exploit a wide range of host instars, such as alteration of host behaviour (Slansky, Reference Slansky1986), manipulation of host development (Vinson & Iwantsch, Reference Vinson and Iwantsch1980), overcoming host immune responses (Strand & Pech, Reference Strand and Pech1995) and egression from hosts to pupate externally to overcome the constraints imposed by hosts on hemolymph-feeding braconids (Strand, Reference Strand, Hochberg and Ives2000; Harvey & Strand, Reference Harvey and Strand2002). We have observed in this study that the larval development in Helicoverpa armigera was usually suspended, but occasionally advanced, in the final instar (table 2), which is similar to the developmental pattern for parasitized host larvae observed in other braconids (Vinson & Iwantsch, Reference Vinson and Iwantsch1980; Tanaka et al., Reference Tanaka, Sato and Hidaka1984; Strand et al., Reference Strand, Johnson and Culin1988). Further studies like ours, examining a range of functional constraints on parasitoid performance in different model systems, will help to elucidate the degree of phenotypic plasticity in the development characteristics of parasitoids attacking hosts with markedly different life histories and ecological requirements (Harvey et al., Reference Harvey, Bezemer, Elzinga and Strand2004).
Acknowledgements
The authors would like to thank Li Na for providing stock insects, Zhu Hongwei for help with statistical analysis and Zhao Lin for assistance with experiments. We are grateful to Kent Daane for his comments on an earlier version of this manuscript, Zhiwei Liu (Biological Science Department, Eastern Illinois University) for valuable views on the revised version and anonymous referees for comments on the manuscript. The study was funded by the National Basic Research and Development Program of China (No. 2002CB111400) and the National Science Fund of China (grant 30570310).
