Classical biological control involves the introduction of an exotic natural enemy to control an introduced pest species. In 1991 the department of Entomology of the Wageningen Agricultural University started a collaborative project with the International Centre for Insect Physiology and Ecology (ICIPE) in Nairobi Kenya, on the biological control of stemborers in Africa. The gregarious endoparasitoid Cotesiaflavipes (Hymenoptera: Braconidae) was chosen as the natural enemy to be introduced against the accidentally introduced pest Chilopartellus (Lepidoptera: Pyralidae) in East-Africa (Overholt et al., 1994; Omwega et al., 1996). The research described in this thesis was related to this project and addressed several aspects of the behavioral ecology of this parasitoid. In the first part, gaps in the knowledge on behavioral ecology of C.flavipes are studied which include the longand short-range searching behavior, some aspects of the life history and host discrimination abilities. The second part focuses on the intraspecific variability in C.flavipes behavior and here we determine to what extent the reported plant and host specificity in C.flavipes has a genetic basis or is due to phenotypic plasticity through learning.Micro-habitat and host location in CotesiaflavipesThe first part addresses the origin of the olfactory stimuli involved in hostmicrohabitat location (chapter 2) and the contact stimuli involved in tunnel and host location on a stemborer infested plant (chapter 3). In chapter 2 it is demonstrated that a major source of the attractive volatiles from the plant-host- complex is the stemborer-injured stem, including the frass produced by the feeding larvae. Moreover, the production of volatiles attractive to a parasitoid is not restricted to the infested stem-part but occurs systemically throughout the plant. The uninfested leaves of a stemborer-infested plant emit volatiles that attract female C.flavipes . An exogenous elicitor of this systemic plant response is situated in the regurgitate of a stemborer larva. When a minor amount of regurgitate is inoculated into the stem of an uninfested plant, the leaves of the treated plant emit volatiles which attract female C.flavipes . Evidence is accumulating that plants are actively involved in attracting natural enemies Dicke 1994). However, whether plants have specifically evolved the ability to release volatiles that attract natural enemies of the herbivore that is attacking them remains a matter of debate (Bruin et al. 1995).Foraging behavior on stemborer infested plantOnce a female C.flavipes has located a stemborer infested plant, it has to locate the concealed host inside the plant stem. In chapter 3 the behavior of female C.flavipes on stemborer infested plants was investigated. It is demonstrated that larval frass, caterpillar regurgitate and holes in the stem are used in host location by C.flavipes . After locating the exit hole of the stemborer tunnel, where larval frass has accumulated, the parasitoid female tries to enter the stemborer tunnel. This can take a long time because the tunnel is often blocked by larval frass and the female sometimes has to squeeze through small holes. The response to host products by C.flavipes seems not to be host species specific. Female C.flavipes respond to frass from four different stemborer species and one leaf feeder. No differences are found in the behavior of C.flavipes on maize plants infested with the suitable host, Chilopartellus (Lepidoptera: Pyralidae), or the unsuitable host, Busseola fusca (Lepidoptera: Noctuidae). Attacking a concealed stemborer larvae in the confined space of stemborer tunnel is not only time consuming but also risky. It is demonstrated that 30-40% of the parasitoids is killed by the spitting and biting stemborer larva. Takasu and Overholt (1996) showed that a female parasitoid has a high probability (0.9) to be bitten to death when it approaches the host towards the head. However, the majority of the females were first able to successfully parasitize its offensive host before being killed. A female C.flavipes needs only a few seconds to inject around 45 eggs into its host. The high probability of mortality at each host encounter results in a very low expectation of the number of lifetime host encounters. The possible consequences of this low lifetime host encounter rate for the evolution of life history- and foraging strategies formed the basis of a large part of this thesis.Life history ofCotesiaflavipesC.flavipes is relatively short lived: without food the parasitoids die within two days, with food and under high humidity conditions they die within 5-6 days. In chapter 3 the fecundity and clutch size allocation of C.flavipes was investigated. It is demonstrated that C.flavipes is pro-ovigenic and has around 150 eggs available for oviposition. In the first encountered hosts 35-45 eggs per host are laid. Thus, a relatively large proportion of the available egg load (20-25%) is allocated to each host, and a female C.flavipes is equipped with an eggload to parasitize 3-4 hosts only. Especially for animals whose lifetime reproductive success is limited by opportunities to reproduce, clutch size theory predicts a maximization of the fitness gain per clutch (Godfray, 1987). This may be true for C. flavipes, which has a short lifespan and a high mortality risk at each host encounter, resulting in a low number of expected lifetime host encounters (Chapter 3). In chapter 4 it is demonstrated that the number of produced adults from superparasitized hosts is equal to that of singly parasitized hosts. This indicates that female C.flavipes indeed lay an optimal clutch size in fourth instar C. partellus larvae.Host- and host-sitediscriminationThe fitness consequences of superparasitism and the mechanism of host discrimination in Cotesia flavipes are described in chapter 4 . Naive females readily superparasitized and treated the already parasitized host as an unparasitized host by allocating the same amount of eggs as in an unparasitized host. However, there was no significant increase in the number of emerging parasitoids from superparasitized hosts due to substantial mortality of parasitoid offspring in superparasitized hosts. Furthermore, the developmental time of the parasitoids in a superparasitized host was significantly longer than in a singly parasitized host and the emerging progeny were significantly smaller (body length and head width). Naive females entered a tunnel in which the host was parasitized 4 hours previously and accepted it for oviposition. Experienced females (oviposition experience in unparasitized host) refused to enter a tunnel with a host parasitized by herself or by another female. In experiments where the tunnel and/or host was manipulated it was demonstrated that the female leaves a mark in the tunnel when she has parasitized a host. The function of the avoidance of superparasitism. in C.flavipes is clear: a discriminating female saves searching time, avoids the wastage of eggs and avoids a direct mortality risk. The mechanism of host discrimination is the recognition of a chemical mark on the tunnel substrate.The role of learning in hostforagingMany studies have shown that parasitoids can learn visual or olfactory stimuli associated with successful host location and use these odours in subsequent foraging decisions (reviewed by Turlings et al., 1993; Vet et al., 1995). The ability to learn has now been demonstrated in more than 20 different parasitoid species and learning in parasitoids seems to be the rule rather than the exception (Turlings et al., 1993). In chapter 5 the role of learning in host foraging in C.flavipes was investigated. Using experimental procedures similar to other parasitoid learning studies, the role of the learning mechanisms priming (i.e. increase in response) and preference-induction in the foraging of C.flavipes was determined. No evidence was found that C.flavipes uses odour learning in hostmicrohabitat location. There was no significant effect of the development and emergence environment on the response level or preference towards infested plant odours. Neither was any evidence found that experience with a particular plant-host-complex during foraging influences subsequent foraging decisions in C.flavipes females.Recent discussions of animal learning emphasize the importance of considering an animals ecology when studying and interpreting its learning abilities. Recently, it has been hypothesized that the adaptive value of learning in foraging is dependent on the predictability of the environment (Stephens, 1993) and the number of lifetime foraging decisions (Roitberg et al., 1993). Learning is not expected when the foraging environment is highly predictable (i.e. the resource is constant) and when animals make only a few decisions while foraging. Taking the ecology of C.flavipes into account it is hypothesized that two factors may be responsible for the lack of learning in foraging in C.flavipes : a predictable foraging environment and the restricted number of lifetime foraging decisions.EVOLUTION OF LIFE HISTORY AND FORAGING STRATEGIESEvolutionary ecological theory concentrates on the interpretation of form and function of individuals as adaptations to their environment. Theories of life history evolution predict what sorts of life history should evolve in specified ecological circumstances (e.g. Steams, 1992; Roff, 1992) and optimal foraging theory addresses the problem of choice among resources or habitats (e.g. Krebs and Davies, 1981; Stephens and Krebs, 1986). It is tempting to relate the ecology of C.flavipes with its life history characteristics and its foraging tactics. The stemborer parasitoid C.flavipes has a peculiar ecology. It not only forages for hosts in a relatively homogeneous and predictable habitat, but it also has a risky attack tactic resulting in a low number of expected lifetime host encounters.The small C.flavipes attacks stemborer larvae by entering the stemborer tunnel (chapter 3). To reach the host, the parasitoid female has to squeeze through small holes in the tunnel which is filled with larval frass. It has been suggested that the dorso-ventral body shape, which is typical of the Cotesia species belonging to the Cotesiaflavipes complex is an adaptation to this ingress behavior (Kimani, pers. comm.). Attacking a host in the confined space of a stemborer tunnel is not without risk for the female parasitoid. At each host attack the female has a considerable risk to be killed by its aggressive host (chapter 3). The short oviposition time (around 40 eggs in 3-4 seconds) may be an adaptation to this mortality risk. The majority of the females that are killed have already successfully parasitized their host.The relatively high mortality risk at each host encounter in combination with the short lifespan results in a very low number of expected lifetime host encounters. This is reflected in the eggload of a female at emergence, which is just enough to parasitize 3-4 hosts (chapter 3). When the probability of surviving to find another host is small, optimal progeny allocation models predicts an optimal 'Lack' clutch size, where fitness is maximized per host (Waage & Godfray, 1985; Godfray 1987). Although it was not tested in depth the results of the superparasitism experiments (chapter 4) indicated that C.flavipes lays an optimal clutch size.The foraging environment of a female C.flavipes can be envisaged as a homogeneous and stable habitat, consisting of a field of perennial grasses with a few prevalent stemborer species. In chapter 5 it is hypothesized that this predictable foraging environment together with the low number of expected lifetime host encounters plays a part in the absence of (odor) learning in C.flavipes host foraging. In chapter 4 it is demonstrated that female C.flavipes leave an external mark on the tunnel substrate after parasitization. It is generally hypothesized that marking evolved as a means for individuals to avoid superparasitizing hosts they themselves previously parasitized (Roitberg and Prokopy, 1987). A female C.flavipes saves time and avoids a superfluous mortality risk by avoiding utilized host tunnels.When a parasitoid has a high mortality risk at each oviposition, life history theory predicts a high selectivity to avoid waste of progeny (Iwasa et al., 1984; Ward, 1992). The parasitoid should not risk her life for low quality hosts, such as unsuitable hosts or already parasitized hosts. However, naive female C.flavipes (no oviposition experience) seem to have a very opportunistic host selection behavior. In chapter 3 it is demonstrated that C.flavipes did attack the (new) unsuitable host B. fusca and in chapter 5 it is found that naive females did utilize a previously parasitized host. The lack of an innate ability or willingness to avoid low quality hosts in C.flavipes may be due to the constrained opportunities to find and parasitize hosts. Each animal faces a evolutionary trade off between reproducing now or in the future, whereby survival chances play a determining role. The best strategy for a recently emerged naive female C.flavipes is to accept the first encountered host, irrespective its quality. Superparasitism pays when future expectancy of host encounter rate is very low. The lower fitness increment of superparasitism (in comparison with single parasitism) will always outweigh the fitness penalty of not finding any unparasitized host.In chapter 5 it is demonstrated, however, that females with oviposition experience do avoid previously utilized stemborer tunnels. The increased choosiness after an oviposition experience in an unparasitized host may be due to the fact that, the parasitoids assessment of host availability has changed. When there is a high chance of finding unparasitized hosts it does pay to reject. Furthermore, in contrast to naive females, oviposition experienced females that superparasitize run the risk to encounter a host they themselves previously parasitized (Van Alphen and Visser, 1990). A safe strategy of females that have already parasitized one or more hosts may be to avoid any already parasitized host to avoid competition among her own progeny.VARIATION IN PARASITOID BEHAVIOR AND BIOLOGICAL CONTROLIn biological control the performance of a released parasitoid population in the field is dependent on the ability of individual females to locate hosts. The behaviour of parasitoids is not fixed and predictable, but most of the times highly variable. This variation in behaviour can be an obstacle in the effective use of natural enemies in biological control, so it is necessary to understand the sources of variation in behavior (Vet et al., 1990; Lewis et al., 1990). In this way we can predict the general behavior of the natural enemy population in the field better and we may even be able to manipulate it.Behavioral variation may exist because individuals differ genetically in propensity to find or accept different hosts. Secondly, individuals may differ because they have experienced different environments (i.e. learning).Local variation in parasitoid behavior and physiologyThe existence of plant specific strains in C.flavipes has been postulated by Mohyuddin and coworkers (e.g. Mohyuddin et al., 1981; Mohyuddin, 1990). However, the genetic and/or phenotypic basis for this reported specificity has never been addressed thoroughly. For instance, early adult conditioning can mimic genetic differences between parasitoid strains. However, in chapter 5 we demonstrated that there is no early adult conditioning for the development and emergence environment in C.flavipes , indicating innate (genetic) differences among strains. Therefore, the between population variation in behavior and physiology of C.flavipes populations was investigated in more detail in chapter 6 . The host selection behavior and physiological compatibility with different stemborers (i.e. parasitoid virulence) was compared for six different geographic strains of C.flavipes that differed in the plant-host-complex they were obtained from. The results of these host selection experiments indicated that there is no interspecific variation in host selection behavior among C.flavipes strains, which contradicts the finding of the Mohyuddin research group. However, the comparative experiments did show variation in reproductive success among strains. The most significant result was that the strain with the longest co- existence time with the new host D. saccharalis , had the highest reproductive success on this host species. It is argued that the earlier reported existence of C.flavipes strains is not based on a differential host selection behavior, but on differences in physiological compatibility between local parasitoid and host population. C.flavipes has beenused on a worldwide scale in biological control against stemborers with varying degree of success (Polaszek and Walker, 1991). The failure of biological control with C flavipes may be due to the introduction of an inappropriate strain. The present study has demonstrated that there is no differential plant preference among strains, but that there are differences in parasitoid virulence among strains. For a reliable biological control with C flavipes it is thus important to select a strain that is physiologically adapted to the target host population.