What are infochemicals?Chemical compounds play an important role in interactions between organisms. Some of these chemicals are to the benefit (e.g. nutrients) or detriment (e.g. toxins) of an organism. Others are of benefit or detriment in an indirect way: through the behavioural response they elicit. The latter chemicals are termed infochemicals (chemicals that, in the natural context, convey information in an interaction between two individuals, evoking in the receiver a behavioural or physiological response that is adaptive to either one of the interactants or both; chapter 2). On an evolutionary time scale, the fate of an infochemical depends on selection pressures on each interactant. Selection pressure is determined by costs and benefits which result from all interactions of an organism in which the infochemical is involved. Yet, for pragmatic reasons, to analyse the function of an infochemical in the biology of an organism, a cost-benefit analysis is made for each interaction between two organisms separately. In this way the cost-benefit analysis is restricted to the smallest number of interactants possible, which ensures its simplicity. Consequently, for each interaction the infochemical is classified according to the corresponding costs and benefits for the two interactants (chapter 2; cf. Nordlund and Lewis, 1976). Moreover, classification also reflects whether the interaction under consideration is between conspecifics or between individuals of different species. This resulted in the terminology represented in Figure 1.1 and Table 1.1 (cf. chapter 2). Its structure and terms are based on those of semiochemicals. However, infochemical terminology differs from semiochemical terminology in two respects (chapter 2):(1) Infochemical terminology regards compounds that convey information, whereas semiochemical terminology in addition also includes toxins (Whittaker and Feeny, 1971; Nordlund and Lewis, 1976; Nordlund, 1981). In some instances toxins or nutrients may convey information. If that is the case, these toxins and nutrients are classified as infochemicals when their role as information carrier is considered. When poisonous or nutritious aspects are considered, they are not classified as infochemicals, but as toxins and nutrients respectively.(2) Semiochemical terminology is based on origin of the compounds, in addition to the cost-benefit analysis. Although knowledge of the origin is Important to understand the interaction between two organisms, it may be very difficult to elucidate the origin (e.g. Brand et al., 1975; chapter 4). Therefore, application of the origin criterion may lead to ambiguities. Because the cost- benefit criterion by itself is good and useful, infochemical terminology is based on that criterion alone.Infochemicals in tritrophic systems.Infochemicals play a role in interactions between consecutive trophic levels (e.g plant-herbivore, phytophagous insect- entomophagous insect; Figure 1.2) (e.g. Nordlund et al., 1981; Visser, 1986). Moreover, infochemicals may also mediate interactions between other trophic levels (e.g. plant-entomophagous insect; Figure 1.2) (Price, 1981). Therefore, to understand the selection pressure on an organism, as a result of an infochemical, all trophic levels involved should be regarded. As a consequence, investigations of infochemicals in interactions between herbivores and their predators should also regard involvement of at least the first trophic level, the plant.The tritrophic system of this study: predatory mites, phytophagous mites and their host plants.The herbivore-predator system investigated most extensively in this thesis consists of phytophagous mites and predatory mites that occur in Dutch orchards. Figure 1.3a,b depicts the two most abundant phytophagous mites that occur as pest organisms in Dutch apple orchards: the apple rust mite, Aculusschlechtendali (Nalepa), and the European red spider mite, Panonychusulmi (Koch) (Van de Vrie, 1973; Van Epenhuijsen, 1981; Gruys, 1982).Several species of predatory mites occur in Dutch orchards. The most abundant of these are Typhlodromuspyri Scheuten (Figure 1.3c), Amblyseiusfinlandicus (Oudemans) and A.potentillae (Garman) (McMurtry & Van de Vrie, 1973; Overmeer, 1981; Gruys, 1982). All three species feed on P.ulmi and A.schlechtendali , as well as on other food sources such as several pollens (Overmeer, 1981; Kropczynska, 1970; Overmeer, 1985).In this system consisting of two phytophagous prey species and three predator species (Figure 1.4a), prey preference of the predators was investigated. Optimal foraging theory predicts that natural selection favours predators preferring prey species that are most profitable in terms of reproductive success (Krebs, 1978). Reproductive success is determined, among others, by development time, oviposition rate, mortality during development and offspring quality. Each of these components can be affected by the prey species consumed. As a first step in analysing which selection pressures may have moulded prey preference of the predatory mites in the system outlined above, I have tested whether prey preference is matched by the associated reproductive success. If this most simple explanation for prey preference does not hold, other explanations should be considered (see below).Do infochemicals play a role in prey preference ?Kairomones (Table 1.1, Figure 1.1) may inform predators on presence and identity of prey (Greany and Hagen, 1981) and thereby affect foraging decisions, such as where to search, how long to search at a specific site, which prey to accept and when to disperse on air currents (chapter 3).Investigation of the response to kairomones may therefore yield information on prey preference. However, the conclusion on prey preference must be restricted to the foraging phase that was studied. Relative costs involved in finding individuals of each prey species might differ for different foraging phases. Therefore, to obtain a comprehensive view of prey preference, several foraging phases should be investigated. Such analyses should be carried out independently to obtain complementary conclusions. In this study, prey preference was determined in three independent analyses.Two laboratory analyses were carried out:- Analysis of response towards volatile kairomones. This investigation regards decisions of the predators when prey individuals are not contacted, as is the situation after termination of aerial dispersal or after eradication of a prey patch.- Analysis of predation rates at different prey supplies. This relates to acceptance/rejection decisions during contacts with prey items.To complement the prey preference analyses carried out in the laboratory, an investigation was made under field conditions: - This was done by determination of diet composition by means of electrophoretic analysis of gut contents of field-collected predators.Spider-mite kairomones in a tritrophic context.Predatory mites distinguish plants infested by spider mites from clean plants by a volatile kairomone (e.g. Sabelis & Van de Baan, 1983). This kairomone seems to be a product of the interaction between plant and spider mites: after removal of spider mites from an infested plant, the plant remains attractive to the predators during several hours, whereas the mites alone do not remain attractive (Sabelis & Van de Baan, 1983; Sabelis et al., 1984a). Current data on spider mite - predatory mite interactions do not explain the role of this infochemical in the biology of the spider mites (cf. chapter 3 for a review). It may, for instance, be an inevitable byproduct of damage inflicted on the plant by the spider mite, and/or have an indispensable function in the biology of the spider mite. Moreover, the plant may be involved in production of the infochemical. To elucidate the role of this volatile infochemical, its effects in interactions between plant and spider mite, between plant and predatory mite and between spider mites of one species should be investigated. Before this can be done, chemical identification of the infochemical is a necessary first step.These investigations were made for a tritrophic system consisting of Lima bean plants, the two-spotted spider mite, Tetranychusurticae Koch and the predatory mite Phytoseiuluspersimilis Athias-Henriot (Figure 1.4b). This system was chosen for practical reasons. The plant and phytophagous mite can be reared throughout the year and therefore, this system is much more suitable to develop a method for the chemical analysis of spider-mite kairomones than a system in which the plant is a perennial.Origin and function of T.urticae kairomone in a tritrophic system.Two-spotted spider mites distinguish between a clean plant and a plant that is infested by conspecifics on the basis of a volatile infochemical (chapter 4). The spider mites move away from heavily infested leaves. This response is advantageous to spider mites on the infested leaf as well as to spider mites that avoid settling on these leaves: increased competition for food is avoided, cf. Wrensch and Young (1978). In addition, the spider mite that disperses thus avoids settling on a spot that has an increased risk of being detected by predatory mites (Sabelis and Van de Baan, 1983). Therefore, the infochemical in this interaction between conspecific spider mites is called a (+,+)dispersing pheromone. Biological evidence suggests that this pheromone is (at least partly) identical to the volatile kairomone to which predatory mites respond (chapter 4).Volatiles emitted from plants infested by T.urticae were identified and subsequent behavioural analyses resulted in identification of four kairomone components that attract the predatory mite P.persimilis : linalool (3,7-dimethyl-1,6-octadiene- 3-ol), methyl salicylate, ( E )-β-ocimene (3,7-dimethyl-1,3( E ),6- octatriene) and 4,8-dimethyl-1,3( E ),7-nonatriene. The structure of these compounds is shown in Figure 1.5. At least two of these (linalool and methyl salicylate) are also components of a kairomone in the interaction between T.urticae and A.potentillae (when reared on V.faba pollen; see below) (chapter 4). Literature data on the behavioural response of T.urticae indicate that one of these kairomone components (linalool) is also a component of the (+,+)dispersing pheromone (Dabrowski and Rodriguez, 1971).All identified kairomone components are well-known in the plant kingdom. This suggests that the plant is involved in production of the infochemical, but it is no proof. It may for Instance be that spider-mite enzymes injected into the plant break down a plant compound. Investigation of e.g. site and moment of production and possible storage of precursors are needed as a next step to elucidate the role of the plant in kairomone production. However, suppose that it is the spider mite who produces the infochemical to serve as a dispersing pheromone. Then, it is not clear why this pheromone should necessarily consist of volatiles. As a result of the production of volatiles the spider mites incur more risks of being detected by predators than by production of non-volatile chemicals. Detection by predators inevitably leads to local extermination of spider mites (Sabelis and Van der Meer, 1986). For this reason it seems more likely that the volatiles are plant produced and that the spider mite makes the best of a bad job by using them as information to decide where not to colonize. To understand the evolution of plant-produced volatiles after herbivore attack, it is crucial to assess how they are produced, how much it costs to produce them and what the benefits are in terms of a lowered probability of herbivore attack.Involvement of volatile kairomones in prey preference of predatory mites.The response of T.pyri and A.potentillae to volatile kairomones is dependent on the diet of the predators. When reared on a carotenoid-poor diet these predators respond to the kairomones of more prey species than when reared on a carotenoid-rich diet (chapters 6, 7 and 8). Carotenoids are indispensable to A.potentillae because of their function in diapause induction (Overmeer, 1985a). The function of these nutrients to T.pyri remains unknown (chapter 8). All prey species to whose kairomones carotenoid-deficient A.potentillae and T.pyri respond can relieve the lack of carotenoids. Carotenoid-containing A.potentillae and T. pyri only respond to the P.ulmi kairomone. The above observations were made for predators that were starved for 20 h. Longer starvation of predators reared on a carotenoid-rich diet also enlarges the number of prey species responded to. Investigations of the response to volatile kairomones indicates that A.potentillae and T.pyri (whether carotenoids are available or not) prefer P.ulmi to A.schlechtendali (chapters 6, 7 and 8) and that A.finlandicus has a reverse preference (chapter 11).This corresponds to conclusions from predation experiments performed at different composition of prey supply (chapters 9 and 11). The observed predation rates when mixed prey supplies were offered, were compared with a model provided with parameters estimated from experiments with each of both prey species alone. Amblyseiuspotentillae and T.pyri fed more on P.ulmi and A.finlandicus fed more on A.schlechtendali than was predicted by the model. This difference between observed and predicted predation rates cannot be explained by a change in behaviour of the prey species as a result of being together, nor by a change in walking behaviour of the predator. Therefore, these data indicate that A. potentillae and T.pyri prefer P.ulmi and that A.finlandicus prefers A.schlechtendali , in terms of a change in acceptance/rejection ratio ('success ratio').Analysis of prey preference under field conditions showed that most T.pyri collected from apple leaves that widely varied in P.ulmi : A.schlechtendali numbers contained P.ulmi esterase, whereas A.schlechtendali esterase was present in a minor fraction of predators (chapter 10). Rust-mite esterase and P.ulmi esterase were found equally frequent in A.finlandicus . The data for A.finlandicus , obtained over a narrower range of prey-number ratios than for T.pyri , do not allow a definite conclusion on prey preference. However, they certainly do not cause rejection of the conclusion on prey preference as obtained in the laboratory analyses (chapter 11). No field data are available for A.potentillae .Because the conclusions on prey preference as determined in these independent analyses are consistent for each predator species, the inference on prey preference is firmly established.Prey preference and reproductive success of predatory mites in an orchard system with two species of phytophagous prey mites.Analysis of reproductive success of these three predator species, when feeding on either P.ulmi or A.schlechtendali , indicates that A.finlandicus selects the best prey species in terms of reproductive success. This predator species suffers high larval mortality on P.ulmi , but not on A.schlechtendali . This results in a much higher intrinsic rate of population increase when feeding on apple rust mites (chapter 12).Amblyseiuspotentillae and T.pyri would also do better by feeding preferentially on A.schlechtendali : development times when feeding on this prey species are shorter than when feeding on P.ulmi , whereas these prey species do not differentially affect mortality or oviposition rate (chapter 12). For A.potentillae this may not be the case at the end of the season because P.ulmi is a better prey species in terms of diapause induction. Thus, on the basis of current data, optimal prey-choice theory cannot satisfactorily predict actual prey peference of A.potentillae and T.pyri . Future investigations should concentrate on e.g. (1) possible effect of competition between prey species on prey availability, (2) possible effect of competition between predator species on prey availability, and (3) possible shift in prey preference during the season.