Host-plant selection by insects a theory based on appropriate/inappropriate landings by pest insects of cruciferous plants

Entomologia Experimentalis et Applicata 96: 91 102, 2000. 2000 Kluwer Academic Publishers. Printed in the Netherlands. 91 Mini review Host-plant selection by insects a theory based on appropriate/inappropriate landings by pest insects of cruciferous plants S. Finch & R.H. Collier Horticulture Research International, Wellesbourne, Warwick CV35 9EF, UK Accepted: April 20, 2000 Key words: Brassica pests, intercropping, undersowing, plant apparency, deterrent chemicals, Resource Concentration Hypothesis, Enemies Hypothesis, Appropriate/inappropriate landing theory Abstract Seven hypotheses, including the Resource Concentration Hypothesis and the Enemies Hypothesis, have been put forward to explain why fewer specialist insects are found on host plants growing in diverse backgrounds than on similar plants growing in bare soil. All seven hypotheses are discussed and discounted, primarily because no one has used any of them to produce a general theory of host plant selection, they still remain as hypotheses. However, we have developed a general theory based on detailed observations of insect behaviour. Our theory is based on the fact that during host plant finding the searching insects land indiscriminately on green objects such as the leaves of host plants (appropriate landings) and non-host plants (inappropriate landings), but avoid landing on brown surfaces, such as soil. The complete system of host plant selection involves a three-link chain of events in which the first link is governed by cues from volatile plant chemicals, the central link by visual stimuli, and the final link by cues from non-volatile plant chemicals. The previously missing central link, which is based on what we have described as appropriate/inappropriate landings, is governed by visual stimuli. Our theory explains why attempts to show that olfaction is the crucial component in the central link of host plant selection proved intractable. The appropriate/inappropriate landings theory is discussed to indicate the type of work needed in future studies to improve our understanding of how intercropping, undersowing and companion planting can be used to optimum effect in crop protection. The new theory is used also to suggest how insect biotypes could develop and to describe why pest insects do not decimate wild host plants growing in natural situations. Introduction Host-plant selection by insects is often divided into host plant finding and host plant acceptance (Thorsteinson, 1960). While the two are easy to separate on paper, the information presented in this short review shows that they are really part of a continuum of three inextricably-bonded links. However, the central link of host plant finding, thought previously to be governed by volatile chemicals, has until now proved intractable to scientific experimentation and it is this area on which we will concentrate. Our review will be based largely on host plant selection by insects associated with cruciferous plants as, since the classical work of Verschaffelt (1910), most theoretical studies on herbaceous plants have used the interaction between insects and cruciferous plants as their test system. Such a selection is not surprising, as cruciferous vegetable and oilseed crops are of high economic importance and are now cultivated on large hectareages in most parts of the world. In addition, cruciferous plants are ideal for biological studies, as their chemistry is well-understood and they support pest species from a wide range of insect Orders (see Finch, 1993). Many researchers have shown that the numbers of pest insects found on cruciferous crop plants are reduced considerably when the background of the crop is allowed to become weedy (Dempster, 1969; Smith, 1969; Dempster & Coaker, 1974; Smith, 1976), when the crop is intercropped with another plant species

92 (O Donnell & Coaker, 1975; Tukahirwa & Coaker, 1982; Ryan et al., 1980; Garcia & Altieri, 1992), or when the crop is undersown with a living mulch (Theunissen & Den Ouden, 1980; Theunissen et al., 1992; Finch & Edmonds, 1994; Theunissen et al., 1995). It has been suggested that when diverse backgrounds disrupt (Vandermeer, 1989) insects from selecting otherwise-acceptable host plants, the action is mediated through (1) physical obstruction (Perrin, 1977), (2) visual camouflage (Smith, 1969; 1976), (3) masking of host plant odours (Tahvanainen & Root, 1972); (4) deterrent ( repellent see later) chemicals (Uvah & Coaker, 1984), or through (5) the non-host plants altering the physiology of the host plants (Theunissen, 1994). Two other suggestions, involving the Resource Concentration Hypothesis (Root, 1973) and the Enemies Hypothesis (Root, 1973), have also been made to explain why fewer phytophagous insects are found on cultivated host plants growing in diverse backgrounds than on similar host plants growing in bare soil. At present, there is a general belief that diverse backgrounds have variable effects on specialist insects (see Andow, 1991; Altieri, 1994) but little, if any, effect on generalist insects (see Risch et al., 1983). In 1983, Risch et al. compared data from 150 published studies to quantify how diverse backgrounds affected specialist insects in annual cropping systems. Their survey revealed that 59% of the species showed a decrease in numbers, 3% an increase, 23% a varied response and 15% no change, when the bare soil in the inter-row spaces was replaced with a diverse background. Unfortunately, the way the data were collected in some studies was questionable and so the above percentage values need to be treated with caution (Coaker, 1990). Nevertheless, even though diverse backgrounds lowered the numbers of about 60% of the 198 species reviewed (Risch et al., 1983), most researchers are still of the opinion that no generic hypothesis will cover all eventualities (see Coaker, 1990). Our initial task in this review will be to discuss the seven hypotheses put forward to date. We will then go on to describe a theory based on appropriate/inappropriate landings which we believe is the key, or missing link, to host plant selection by phytophagous insects. Finally, we will use the new theory (1) to discuss the type of information required to make intercropping, undersowing and companion planting more successful, (2) to suggest how insect biotypes could develop, and (3) to describe why wild host plants are not decimated by pest insects. This review will be critical rather than encyclopaedic (see Altieri, 1994), as we would like to remove, rather than perpetuate, many of the myths and untested assumptions found currently in the literature. Description and discussion of the seven earlier hypotheses 1. Physical obstruction. This hypothesis was used to describe those situations in which the host plants were in effect hidden physically by using larger or taller non-host plants. For example, Altieri & Doll (1978) used tall maize plants to protect bean plants from pest infestations. Tall plants were considered to be effective because they obstructed the movement of the pest insect within the cropping system (Perrin, 1977). It could be argued that there is an element of physical obstruction when clover disrupts host plant finding by pest insects of brassica crops (Theunissen et al., 1992) as, to have maximum impact, the foliage of the clover has to surround much of the host plant (Finch & Edmonds, 1994; Theunissen, 1994). Although clover growing in such close proximity to the host plant will obviously obstruct the searching insects, no suggestions have been made of how this mechanism might operate. However, a mechanism relying solely on physical obstruction is not supported by the recent findings of Finch & Kienegger (1997). These authors used clover plants, which they desiccated so that the clover plants retained the same architecture as the living plants and differed only by being brown rather than green. The numbers of eggs laid when the cabbage root fly (Delia radicum), the diamond-back moth (Plutella xylostella), and the large white butterfly (Pieris brassicae) were presented with host plants surrounded by desiccated (brown) clover did not differ from the numbers laid on host plants presented in bare soil. Hence, the physical presence of the clover was not sufficient on its own to reduce the numbers of eggs laid, a reduction occurred only when the surrounding clover was green (Finch & Kienegger, 1997). 2. Visual camouflage. This hypothesis was based on the two types of visual stimuli that induce low-flying insects to land on plants. The first is a directed response to the colour of the plant (Moericke, 1952), which in most cases means green, and the second is an optomotor response in which landing is provoked by plants looming up along the path of the flying insect

93 (Kennedy et al., 1961). Anything that competes with such stimuli, such as other green plants, or raising the height of the overall background with weeds (Smith, 1976), so that the distance over which the host plant can be separated from the background is foreshortened, helps to camouflage visually the host plants. This makes the host plants less apparent (Feeny, 1976) amongst the foliage of the non-host plants. In 1976, Smith showed that clean-weeding of Brussels sprouts (Brassica oleracea var. gemmifera), the normal agricultural practice, provided ideal conditions for colonization by aphids, whitefly and certain Lepidoptera. Although she showed clearly that the cabbage aphid (Brevicoryne brassicae), preferred plants which stood out against a background of bare soil, she made no attempt to determine how such a mechanism might operate. 3. Masking of host plant odours. The release of odour-masking substances into the air by non-host plant species is considered to confer some protection to the associated host plants. This associational resistance, as it was termed by Tahvanainen & Root (1972), has been reported also by Buranday & Raros (1975) and Perrin & Phillips (1978). Although this hypothesis seems plausible, few data have been collected during the last 25 years to support it. The possibility that the odour of the host plant could be masked by that of the non-host plant (Tahvanainen & Root, 1972) now seems much less likely, though not impossible (Thiery & Visser, 1986). For example, host plant selection by the cabbage root fly was disrupted when its host plants were surrounded by a range of different plants including weeds (Dempster, 1969; Smith, 1976), spurrey (Spergula arvensis)(theunissen & Den Ouden, 1980), peas (Pisum sativum) (Kostal & Finch, 1994), rye-grass (Lolium perenne) (Kostal & Finch, 1994), or clover (Theunissen et al., 1992; Finch & Edmonds, 1994). As each of these non-host plants has a different odour profile, it seems highly improbable that they would all be capable of preventing an adapted specialist insect from finding its host plants. Furthermore, observations made in windtunnels revealed that brassica plants growing in clover were approached by cabbage root flies as readily as brassica plants growing in bare soil, indicating that the odours from the clover did not mask those of the flies host plants (Tukahirwa & Coaker, 1982). More striking, however, was that the same disruptive effect could be produced by surrounding the host plants with plant models made from green card (Kostal & Finch, 1994), or simply with sheets of green paper (Ryan et al., 1980; Kostal & Finch, 1994), neither of which were releasing plant chemicals. It seems that once the characteristic host plant chemicals stimulate the insects to land, the disruption is caused simply by providing the insects with a greater number of green surfaces on which to land. 4. Repellent chemicals. It is implicit in this hypothesis (Uvah & Coaker, 1984) that the odours given off by the non-host plants are sufficiently strong to actually repel (Dethier et al., 1960) the searching insects. It was suggested that the diamond-back moth could be repelled from cabbages by intercropping the cabbages with tomatoes (Lycopersicon esculenta) (Buranday & Raros, 1975) and that the highly odorous ragweed (Ambrosia artemisifolia) couldbeusedtorepelthe cabbage flea beetle (Phyllotreta cruciferae) from crops of collard (Brassica oleracea var. acephala) (Tahvanainen & Root, 1972). Such suggestions were made to describe why pest insect numbers were different in the two situations. They were not based on scientific experimentation. Whether or not true deterrency is a mechanism still needs to be proven. Deterrency usually involves highly aromatic plants that often have to be crushed and tested in small confined spaces in the laboratory to show that they are actually capable of repelling pest insects. As such tests are far from natural, the validity of using such data during the synthesis of new behavioural mechanisms is questionable. In reality, no experimental evidence has been produced during the last 15 years to support the hypothesis (Uvah & Coaker, 1984) that the non-host plants produced their effects through chemical deterrence. Plants chosen for their odorous nature, such as French marigolds (Tagetes patula), failed to deter the carrot fly (Psila rosae) whenusedastheintercrop in carrots (Uvah & Coaker, 1984). In addition, oviposition by the diamond-back moth was similar on Brussels sprouts plants intercropped with plants of sage (Salvia officinalis L.) and thyme (Thymus vulgaris L.) (Dover, 1986), two plant species selected for their pungent odours. Extracts of the essential oils of sage and thyme were shown to reduce oviposition by the diamond-back moth but the effect resulted from contact stimuli and not from repellent volatile stimuli (Dover, 1985). Doubtless, a wide range of contact chemicals play a major role during host plant acceptance (Schoonhoven, 1968), but as these come into play only after an insect has landed, they are included only in the second part of this review, which is con-

94 cerned with Link 3 (see later), the part involving host plant acceptance. 5. Altering the profiles of the host plant odours. While this seems a novel mechanism, it relies upon the host plants being unable to metabolize certain of the chemicals they take up from the soil, so that such chemicals in effect change the subsequent physiology of the plant. Many claims are made that African marigolds (Tagetes spp.) planted between rows of crop plants reduce pest numbers. It is clear from the earlier discussions that this is unlikely to be a direct effect of the odours of the African marigolds repelling the colonizing insects. However, it is well-known that species of African marigolds release large amounts of root exudates which can be taken up by adjacent plants (Rovira, 1969). It is possible, therefore, that any host plant growing in an intercrop could be affected directly by chemicals taken up through its roots (Theunissen, 1994) rather than by having its odour masked. To prevent this happening, the test plants in the experiments of Finch & Kienegger (1997) were left in their pots throughout the test periods. As the effects of the clover were still evident and often within minutes of starting an experiment (Finch & Kienegger, 1997), a generic mechanism based on the non-host plants (here clover) causing physiological changes in the host plants (Theunissen, 1994) cannot be supported. 6. The Resource Concentration Hypothesis. The last two hypotheses (6 & 7) are the ones quoted most frequently. Both were derived from one study in which Root (1973) monitored insect distribution during three field seasons in pure stands of collard plants and in single rows of collard plants bounded on each side by diverse meadow vegetation. The Resource Concentration Hypothesis states that phytophagous insects are more likely to find and remain on host plants that are growing in dense or nearly pure stands (Root, 1973). Phytophagous insects that arrive in a clump of host plants, by whatever means, and find conditions suitable, will tend to remain in the area. This trapping effect of host patches will depend upon several factors such as the size and the purity of the plant stand (Pimentel, 1961) and the type of host plant required by the phytophagous insect. In many cases this accumulation of specialist insects on a concentrated resource (here cultivated Brassica plants) will be sufficient to increase the numbers of phytophagous insects in that locality. Again this hypothesis describes simply the effect of changes in the Figure 1. Numbers of eggs laid on brassica plants presented in backgrounds of subterraneum clover (Trifolium subterraneum) expressed as a percentage of the numbers of eggs laid on brassica plants presented in bare soil. Cabbage aphid shown as numbers of aphids rather than numbers of eggs (From Finch & Kienegger, 1997). purity of a host plant stand on insect numbers. The author (Root, 1973) makes no attempt to develop a general theory. 7. The Enemies Hypothesis. Contrary to five of the earlier hypotheses which claim that the differences are due to the direct effects of the diverse backgrounds on the behaviour of the pest insects, this hypothesis, like hypothesis 5, proposes that the effects are indirect. In essence, this hypothesis proposes that lower numbers of phytophagous insects are found in complex environments because predators and parasitoids are more effective in such situations (Root, 1973). Thus outbreaks of phytophagous insects are checked early by the higher numbers of enemies that can be supported by the diverse resources available in complex environments. Unfortunately, Root (1973) found that the effectiveness of enemies did not differ significantly between collards grown as pure stands and those grown in single rows amongst diverse meadow

95 vegetation. Nevertheless, Root pursued this enemies hypothesis by discussing sets of data collected mainly in England (see Root, 1973). His own, more extensive, data, however, indicated clearly that the diversity of both predators and parasitoids was higher in the pure stands, probably because more prey/host species were also present in that habitat. Consequently, he concluded that factors other than natural enemies were responsible for much of the differences in insect numbers recorded between simple and diverse habitats. Although Root (1973) himself discounted the enemies hypothesis, many subsequent workers have championed the cause of predators, often on the flimsiest of evidence or without collecting the data necessary to support their claims. Conclusions from recent work Although authors have indicated that diverse backgrounds can affect host plant selection in the seven ways described above, it is hard from the results presented in recent publications (Kostal & Finch, 1994; Finch & Kienegger, 1997) to refute the more simplistic view that one mechanism is operating against all species. The work done by Finch & Kienegger (1997) included experiments on the cabbage moth (Mamestra brassicae), the diamond-back moth, the garden pebble moth (Evergestis forficalis), the small white butterfly (Pieris rapae), the large white butterfly, the cabbage aphid, the cabbage root fly, and the mustard beetle (Phaedon cochleariae). Despite these eight test species being from four different insect Orders, the ability of each of them to find host plants was affected adversely, though to differing degrees, when their host plants were surrounded by clover (Figure 1). It appeared that differences in the initial rates of colonization was the factor that regulated the numbers of phytophagous insects found on host plants growing in bare-soil or clover (Finch & Edmonds, 1994), as differences between the two situations were often apparent within minutes of starting an experiment (Finch & Kienegger, 1997). Description and discussion of the new theory Unfortunately, no one has yet developed any of the earlier hypotheses into a robust general theory, they are still only hypotheses. Hence, we have generated our own theory from detailed studies of insect behaviour. Instead of the seven hypotheses described previously, we believe that a mechanism that we have described as appropriate/inappropriatelandings (Finch, 1996) is the central link in host plant selection by insects. In the overall system, the new theory of host plant selection can be divided into a chain of actions involving just three inextricably-bound links. In the first link, volatile chemicals emanating from plants indicate to flying receptive insects that they are passing over suitable host plants (Figure 2-1). The primary action of these volatile chemicals is to stimulate the insects to land (Figure 2-2). Under suitable weather conditions, these chemicals may also provide some directional information (Finch & Skinner, 1982), but this is of secondary importance. The tenet is that the amounts of volatile chemical released from plants that impinge upon the insect s receptors are sufficient to arrest, but rarely sufficient to provide accurate directional information to flying phytophagous insects. Insects that fly over plants growing in bare soil will be stimulated to land on host plants, the only green objects available to them (Figure 2-3a), as most phytophagous insects avoid landing on brown surfaces, such as soil (Kostal & Finch, 1994). When host plants are growing in bare soil, most landings will be what we have classed as appropriate and so the host plants will in effect concentrate (Root, 1973) the insects. In contrast, insects flying over host plants surrounded by clover land in proportion to the relative areas occupied by leaves of the host (Figure 2-3a) and non-host (Figure 2-3b) plants, as specialist phytophagous insects do not discriminate between the two when both are green (Moericke, 1952; Prokopy & Owen, 1983; Prokopy et al., 1983; Kostal & Finch, 1994). Hence, any landings made on the non-host plant, here represented by clover (Figure 2-3b), are classed as inappropriate. The amount of time the insects spend on the leaves of the non-host plants before taking off again is governed by whether the insects receive acceptable or antagonistic stimuli through their tarsal receptors (see Figure 4). Once the insects are again airborne (Figure 2-4), if they are stimulated to land after flying only a relatively short distance (Figures 2-5 and 6), they could land on a host plant. In all situations, however, the plant on which the insect first lands, even if it is a host plant, may not stimulate the insect sufficiently, via its contact chemoreceptors on the tarsae or head appendages, to arrest it, and the overall process will be repeated. If this represented the complete system, then under nochoice situations in the field, it could just be a matter of time before the numbers of eggs laid on host plants

96 Figure 2. Schematic diagram to illustrate how diverse backgrounds, here represented by clover (Trifolium spp.), influence host plant finding by the cabbage root fly. Numbers represent insect actions 1 7 (see text). Figure 3. Schematic diagram to illustrate how diverse backgrounds, here represented by clover (Trifolium spp.), influence host plant acceptance by the cabbage root fly. Numbers represent the four (mean no.) leaf-to-leaf flights made by the fly to ascertain whether the plant is a suitable substrate around which to lay its eggs.

97 growing in diverse backgrounds were similar to those laid on host plants growing in bare soil. However, this does not occur, as there is a second phase to host plant finding. This second phase can be illustrated (Figure 3) most clearly by data collected from a detailed study of the cabbage root fly. The figure shows that before accepting a host plant as a suitable site for oviposition, receptive female cabbage root flies make, on average, four spiral flights before laying alongside the plant (Kostal & Finch, 1994). Hence, the insects stand a much greater chance of losing the host plant in a diverse background as, on average, they repeat the initial appropriate/inappropriate landing procedure a further three times (Kostal & Finch, 1994. Observations under laboratory conditions showed that for every 100 females that landed on a brassica plant surrounded by bare soil, 36 (Figure 3a) received sufficient stimulation from the plant to be induced to lay eggs. In contrast, only seven (Figure 3b) out of 100 females that landed on host plants surrounded by clover managed to lay eggs. Fewer flies managed to lay in this situation, because following each short spiral flight, a proportion of the flies landed on the leaves of the surrounding clover plants. This failure to re-contact a leaf of a host plant after any spiral flight prevented the females from accumulating, within the allotted time, sufficient stimulation from the host plant to be induced to lay. Hence, the barrier that this fly faces when its host plants are grown in diverse backgrounds is not chemical (O Donnell & Coaker, 1975; Dover, 1986) nor mechanical (Theunissen & Schellling, 1996) but behavioural, simply because during the innate series of spiral flights the fly must continue to accumulate more positive host plant stimuli each time it lands. The amount of stimulation the female picks up on each landing is crucial and this is where the phase of host plant finding (Link 2) becomes truly integrated with host plant acceptance (Link 3). In essence, the complete system really involves finding and re-finding the host plant. The schematic representation shown in Figure 4 indicates that the female cabbage root fly may only have to visit two leaves of a highly stimulating plant [1] compared to six leaves on a poorlystimulating plant [3] before finding it an acceptable site on which to lay eggs. Other insects, however, may accumulate sufficient stimuli to keep them searching [4], but not sufficient stimuli to induce oviposition and so will fly away. A similar outcome results when insects visit several leaves (here shown as 3) but do not manage to accumulate sufficient stimuli in the allotted time to be induced to stay [5]. Two other variations occur when the insects land initially on a stimulating leaf but subsequently on a non-stimulating leaf. It does not matter whether this leaf is from a host [6] or a nonhost plant [7], as anything that interrupts the rate of accumulation of positive stimuli causes the insect both to abort its attempt to lay and to move elsewhere. In addition, interspecific competition may also become important. At any stage during the host plant selection process, many of the new immigrants may not remain on otherwise-acceptable plants if they are colonized already by certain, but not all, of the other insect species present in the pest complex (Jones & Finch, 1987). The physiological status of the insect, which depends partly on its age and also on how long it has been deprived a suitable oviposition site, also has to be superimposed upon this already complex system (Barton Browne, 1993). With time, phytophagous insects tend to become less discriminating in their choice of oviposition sites (Kostal & Finch, 1996). The condition of the plant is also extremely important, as some cruciferous plant species are more highly-preferred than others and during their phase of exponential growth many individual plants become highly stimulating to insects (see Finch, 1980). However, even when the insect and the plant are both in the appropriate physiological state, it counts for nothing the moment the insect makes a wrong choice and alights on any green object other than a leaf of a host plant. This, however, is tempered by the fact that when the host plant is highly stimulating the insect has to visit fewer leaves and so has less chance of making an inappropriate landing. In addition, the highly-stimulating plants invariably induce the individual insects to lay more eggs (Roitberg et al., 1999). Detailed descriptions of the multitude of other factors involved during host plant acceptance can be found in many of the papers published recently in the proceedings of the Tenth International Symposium on Insect-Plant Relationships (Simpson et al., 1999). Although the other seven test species (see Figure 1 for names) have not been studied in detail, records of the movements of the small white butterfly in the field showed that it always made contact with several leaves, interspersed with short flights, before laying an egg (Jones, 1977). Jones concluded that it needed time to get the next egg ready, but these short flights could represent a behavioural repertoire similar to that of the cabbage root fly. It seems likely, therefore, that the relative differences in the effect that

98 Figure 4. The number of leaf landings a cabbage root fly may have to make before accepting a plant as a suitable ovipostion site or deciding to fly elsewhere. The numbers in [] represent seven possible variations in the pattern of insect behaviour. diverse backgrounds have on host plant selection by the test species (Figure1) may simply reflect the numbers of contacts/re-contacts the insect has to make to accumulate sufficient positive stimuli to lay. Results from recent experiments (Figure 1) indicated that the diamond-back moth was the species affected least by diverse backgrounds (Finch & Kienegger, 1997). It raises the question of whether the diamond-back moth has become the major world pest of cruciferous crops simply because it has a limited behavioural repertoire prior to oviposition (Harcourt, 1957) and so lays on more or less the first host plant leaf it encounters. General discussion The appropriate/inappropriate landing theory can be used to explain why certain aspects of host plant finding by phytophagous insects, supposedly regulated by volatile plant chemicals, proved intractable in the past. Compared to our simple theory of host plant finding, extremely complex theories had to be generated to suggest how volatile chemicals could guide phytophagous insects to their host plants. For example, it is known that the odorous environment of a given insect is a kaleidoscopically shifting maze of overlapping active spaces (Wilson, 1970). Therefore, it was suggested that it is in this shifting maze that the insect must find the active space containing those few signals that will lead it to its host plant. However, in the open air, it is turbulence rather than diffusion that determines the distribution of odorous molecules (Wright, 1958; Bossert & Wilson, 1963), and there is nearly always a prevailing wind to blow the odorous molecules away from the plant. Registering directional cues from odorous molecules is even more complicated for flying insects, as the movement of the air relative to the insect depends on the insect s own activity, and wind direction has to be registered in the absence of fixed markers (Finch, 1980). The evidence that volatile chemicals are the main regulatory stimuli in the central link of host plant finding is weak, as the maximum distance recorded for insect orientation to host plant volatiles in the field is only a few metres (Hawkes, 1974; Finch, 1980). In wind tunnel experiments, cabbage root flies flew upwind to a cruciferous plant odour released at 2.5 g day 1 (Hawkes & Coaker, 1979), a rate similar to that dispensed from field traps (Finch, 1980). This rate of release is at least 10 5 times higher than the

99 amount of chemical released from a healthy cruciferous plant (Finch, 1980). Although the flies moved upwind in response to this odour, less than 10% of the flights lasted more than 0.5 m. While the shortness of such flights was described as unexpected (Hawkes et al., 1978), it would not have been unexpected had the chemicals involved been regarded as arrestants rather than attractants (see Dethier et al., 1960; Kennedy, 1978). Similarly, in insect traps releasing large amounts of chemical to provide directional cues in the field, many insects missed the trap (Finch & Skinner, 1982) and failed to enter subsequently, suggesting again that the chemicals were arrestants rather than true attractants. The current mechanism seems much more robust than one based on volatile chemicals, as the sooner an insect lands, the sooner it is freed from the plethora of problems it faces while still in the air. The appropriate/inappropriate landing theory works equally well for generalist feeders, where the decision of whether to stay is determined primarily by the chemicals the insect detects via its contact chemoreceptors once it has landed on a leaf. However, it should be remembered that if an insect that is regarded as a generalist has been stimulated to land by volatile chemicals released from a specific plant species, the insect may continue its search for such a plant rather than accepting the first suitable plant it encounters. Finally, the appropriate/inappropriate landing theory appears to apply equally well to nocturnal insects, as oviposition by the diamond-back moth and the cabbage moth was not disrupted when their host plants were surrounded by brown clover (Finch & Kienegger, 1997). Although both of these moth species are considered to be nocturnal, much of their egg-laying activity is concentrated at, or shortly before, dusk. The results of Finch & Kienegger (1997) indicate that during this period both species of moth appeared to be able to discriminate between green and brown objects. From a crop protection standpoint, the more nonhost plants removed from any crop area, the greater chance an insect has of finding a host plant. Hence, our current cultural methods are exacerbating our pest control problems, as bare-soil cultivation ensures that crop plants are exposed to the maximum pest insect attack possible in any given locality. Future work If the theory based on appropriate/inappropriate landing is accepted it raises searching questions regarding several aspects of entomological research. In the first instance, we have expressed considerable doubts about whether host plant volatile chemicals are truly attractants or simply arrestants for receptive insects (Kennedy, 1978) a debate that has rumbled on for years. Most of the detailed experiments with host plant volatile chemicals (e.g., Hawkes & Coaker, 1979) have been done by releasing plant odours from a point source sited at one end of a windtunnel, introducing supposedly responsive insects and then recording whether the insects move upwind to the source of the odour. As we indicated earlier, the results from this approach have been disappointing (see Hawkes, 1974; Finch, 1980). Often the only way to obtain data is to place a visual stimulus, normally a green coloured object, alongside the site from which the volatile chemical is being released (see Rojas & Wyatt, 1999). To prove that the odours are arrestants rather than attractants, the set-up within the wind tunnel would have to be modified considerably. We envisage that the flow of odour-free air through the wind tunnel would need to be balanced carefully to keep receptive insects flying above green circular objects. If a pulse of odour then stimulated the insects to land on the green objects, it should help to prove that the volatile chemicals behave primarily as arrestants, as the odour source and visual stimulus could in some of the larger wind tunnels be 2 3 m apart. Provided the insects could be kept flying in this way, and some aphids can (Kennedy & Thomas, 1974), this might also be a way of determining the period of flight aphids require before they become receptive to volatile host plant stimuli. This could be done by releasing pulses of a volatile stimulus at known time intervals and using odour-free air during the rest of the insects s flight period. If, as the appropriate/inappropriate landing theory suggests, it is just the number of green objects surrounding a host plant that reduces colonization by pest insects, then it should not be too difficult in future to quantify the type of plant architecture needed for the diverse background to reduce pest insect numbers in any given crop. If several plant types proved appropriate for a given cultivated crop, it would then simply be a case of choosing the one that caused the least reduction in yield to the harvested product.

100 There is also a need to obtain a better understanding of companion planting (Franck, 1983), a practice used frequently by organic growers. The earlier data showed that there is no scientific evidence that the odours from highly aromatic plants can actually deter pest insects (Dover, 1985). This, therefore, brings into question how these aromatic plants produce their effects. A survey of the literature should help to show whether such systems are effective only when the companion plants possess relatively large amounts of foliage when compared to the crop plants. If this can be substantiated, the differences recorded may simply be a reflection of appropriate/inappropriate landings and again have little to do with volatile chemicals, no matter how pungent the plant odours might appear to the researcher. Proving this should not be too difficult, as one of the frequently-quoted examples of companion planting describes how growing onions helps to reduce carrot fly (Psila rosae) infestations (Uvah & Coaker, 1984). At the stage of growth when the carrots need protecting from the fly, onions are little more than narrow, vertical, cylindrical green objects. Hence, surrounding carrot plants with known numbers of odorous onion plants or comparable numbers of non odorous green sticks (the onion mimics) should allow the contribution made to companion planting by plant volatile chemicals to be quantified. Apart from its impact in practical pest control situations (see Cromartie, 1981), the appropriate/inappropriate landing theory helps also to explain why wild host plants growing amongst other plants in natural vegetation are rarely decimated by pest insects. However, a certain proportion of the insects do develop on wild-host plants. The question then is do such insects prefer to remain on the wild host plants in subsequent generations, a situation that could give rise to biotypes? The answer to this question is of considerable practical importance. Future work is required, to determine what proportion of any given pest population develops on wild-host plants and whether such insects readily switch back to the cultivated crop plants. If they do not, then it should be possible to control certain pest insects by isolating new crops from earlier infestations. Additional work is required also to determine whether appropriate/inappropriate landings influence the overall distribution of the parasitoids of the pest species. It is well-documented that such insects use both host plant chemicals and kairomones to locate their host insects. The simple experiment of Richards (1940), in which he placed cabbage plants infested with small white butterfly larvae in bare-soil crop situations and in hedgerows, showed clearly that one specific parasitoid (Apanteles rubecula) was also affected adversely when the plants on which its host insects were feeding were placed into a diverse background. Further work is needed on other parasitoids to determine whether this is a general phenomenon. If it is, then the suggestions of some researchers that diverse background have adverse effects on pest insects and no effect on their parasitoids warrant further study. Similarly, with respect to the Enemies Hypothesis (Root, 1973), by placing plants infested with pest insects into bare soil and diverse backgrounds it should be relatively easy to show whether predation is higher on infested plants surrounded by non-host plants than on plants surrounded by bare soil. It should also be easy under field conditions using mark/recapture methods (see Southwood, 1978) to show whether released predators aggregate mainly around brassica plants growing in bare soil or around similar plants growing in diverse backgrounds. As much of the work in this review has been based on detailed information on the cabbage root fly, work is needed to determine the detailed activity of each of the other pest species. It appears that diverse backgrounds may have a greater effect on the cabbage root fly than on the diamond-back moth simply because, before acquiring sufficient stimulus to lay, the cabbage root fly repeats the sequence of host plant finding four times, but the diamond-back moth only once. It would be difficult to prove that once the cabbage root fly starts one of its spiral flights it is again stimulated to land (arrested) by volatile chemicals, as the distances between taking-off and landing again are extremely short. However, olfaction is considered to be used by phytophagous caterpillars in choosing their feeding sites (Chapman, 1999) and so the intervention of olfactory stimuli again at this stage cannot be ruled out. Apart from differences between pest species, studies are needed on the variation between the individuals within a population of a particular pest and how this is expressed on an individual plant basis both within a specific cultivar and on a range of individual plants from different cultivars. In essence, this encompasses both the plant preference aspect of field trials to select partially-resistant crop plants and the way to select highly-stimulating plants for use as trap crops. The ideal trap crop could be one from which the insect obtains the entire stimulus it needs to lay from the first trap-crop leaf it encounters.

101 Irrespective of how this theory is received by researchers studying insect/host plant attractants, appropriate/inappropriate landings, does appear to provide a robust description of host plant selection by insects under a wide range of different conditions. It shows how elements of all the earlier hypotheses can be incorporated into the overall system, and in particular how the lack of detailed research on insect behaviour, and in particular on visual stimuli, has led to many of our current anomalies. Apart from the future work proposed above, it will be interesting to determine whether the same theory regulates host plant selection by specialist insects found associated with plant families other than the Cruciferae. While we believe the simplicity of our theory makes it allembracing, only time will tell whether our optimism is justified. Acknowledgements We thank the UK Ministry of Agriculture, Fisheries and Food (Contact: Dr Sue Popple) for supporting this work as part of Project HH1815SFV. We thank also Dr Vladimir Kostal, Czech Academy of Sciences, Branišovská 31, 370 05 České Budějovice, Czech Republic and Miss Manuela Kienegger, Ludwig- Boltzmann Institute, Vienna, Austria for the help with the early experiments during their respective study periods at HRI-Wellesbourne. References Altieri, M. A., 1994. Biodiversity and Pest Management in Agroecosystems. Haworth Press Inc., New York, 185 pp. Altieri, M. A. & J. D. Doll, 1978. Some limitations of weed biocontrol in tropical ecosystems in Columbia. In: T.E. Freeman (ed.), Proceedings IV International Symposium on Biological Control of Weeds. University of Florida, Gainesville, pp. 74 82. Andow, D. A., 1991. Vegetational diversity and arthropod population response. Annual Review of Entomology 36: 561 586. Barton Browne, L., 1993. Physiologically induced changes in resource oriented behavior. Annual Review of Entomology 38: 1 25. Bossert, W. H. & E. O. Wilson, 1963. The analysis of olfactory communication among animals. Journal of Theoretical Biology 5: 443 469. Buranday, R. P. & R. S. Raros, 1975. Effects of cabbage-tomato intercropping on the incidence and oviposition of the diamondback moth, Plutella xylostella (L.) Phillippine Entomology 2: 369 375. Chapman, R. F., 1999. It s all in the neurones. Entomologia Experimentalis et Applicata 91: 259 265. Coaker, T. H., 1990. Intercropping for pest control. In: BCPC Monograph No. 45. Organic and Low Input Agriculture, pp. 71 76. Cromartie, W. J. Jr., 1981. The environmental control of insects using crop diversity. In: D. Pimentel (ed.), CRC Handbook of Pest Management in Agriculture, Volume 1. CRC Press, Boca Raton, Florida, pp. 223 251. Dempster, J. P., 1969. Some effects of weed control on the numbers of the small cabbage white (Pieris rapae L.) on Brussels sprouts. Journal of Applied Ecology 6: 339 345. Dempster, J. P. & T. H. Coaker, 1974. Diversification of crop ecosystems as a means of controlling pests. In: D. Price Jones & M. E. Solomon (eds), Biology in Pest and Disease Control. John Wiley & Sons, New York, pp. 106 114. Dethier, V. G., Browne, L. B. & C. N. Smith, 1960. The designation of chemicals in terms of the responses they elicit from insects. Journal of Economic Entomology 53: 134 136. Dover, J. W., 1985. The responses of some Lepidoptera to labiate herb and white clover extracts. Entomologia Experimentalis et Applicata 39: 177 182. Dover, J. W., 1986. The effect of labiate herbs and white clover on Plutella xylostella oviposition. Entomologia Experimentalis et Applicata 42: 243 247. Feeny, P. P., 1976. Plant apparency and chemical defence. In: J. Wallace & R. Mansell (eds), Biochemical Interactions Between Plants and Insects. Recent Advances in Phytochemistry 10: 1 40. Finch, S., 1980. Chemical attraction of plant-feeding insects to plants. In: T. H. Coaker (ed), Applied Biology V. Academic Press, London & New York, pp. 67 143. Finch, S., 1993. Integrated pest management of the cabbage root fly and the carrot fly. Crop Protection 12: 423 430. Finch, S., 1996. Appropriate/inappropriate landings, a mechanism for describing how undersowing with clover affects host plant selection by pest insects of brassica crops. IOBC/WPRS Bulletin 19 (11): 102 106. Finch, S. & G. H. Edmonds, 1994. Undersowing cabbage crops with clover the effects on pest insects, ground beetles and crop yield. IOBC/WPRS Bulletin 17 (8): 159 167. Finch, S. & M. Kienegger, 1997. A behavioural study to help clarify how undersowing with clover affects host plant selection by pest insects of brassica crops. Entomologia Experimentalis et Applicata 84: 165 172. Finch, S. & G. Skinner, 1982. Trapping female cabbage root flies (Delia radicum (L.)) (Diptera: Anthomyiidae) with allylisothiocyanate-baited traps. Bulletin of Entomological Research 72: 165 173. Franck, G., 1983. Companion Planting Successful Gardening the Organic Way. Thorsons Publishers Limited, Wellingborough, England, 128 pp. Garcia, M. A. & M. A. Altieri, 1992. Explaining differences in flea beetle Phyllotreta cruciferae Goeze densities in simple and mixed broccoli cropping systems as a function if individual behavior. Entomologia Experimentalis et Applicata 62: 201 209. Harcourt, D. G., 1957. Biology of the diamondback moth, Plutella maculipennis (Curt.) (Lepidoptera: Plutellidae) in Eastern Ontario. II Life-history, behavior, and host relationship. Canadian Entomologist 89: 554 564. Hawkes, C., 1974. Dispersal of adult cabbage root fly (Erioischia brassicae (Bouché) in relation to a brassica crop. Journal of Applied Ecology 11: 83 93. Hawkes, C. & T. H. Coaker, 1979. Factors affecting the behavioural responses of the adult cabbage root fly, Delia brassicae, to host plant odour. Entomologia Experimentalis et Applicata 25: 45 58. Hawkes, C., Patton, S. & T. H. Coaker, 1978. Mechanisms of host plant finding in adult cabbage root fly, Delia brassicae. Entomologia Experimentalis et Applicata 24: 419 427.

102 Jones, R. E., 1977. Movement patterns and egg distribution in cabbage butterflies. Journal of Animal Ecology 46: 195 212. Jones, T. H. & S. Finch, 1987. The effect of a chemical deterrent, released from frass of caterpillars of the garden pebble moth, on cabbage root fly oviposition. Entomologia Experimentalis et Applicata 45: 283 288. Kennedy, J. S., 1978. The concepts of olfactory arrestment and attraction. Physiological Entomology 3: 77 89. Kennedy, J. S. & A. A. G. Thomas, 1974. Behaviour of some lowflying aphids in wind. Annals of Applied Biology 76: 143 159. Kennedy, J. S., Booth, C. O. & W. J. S. Kershaw, 1961. Host finding by aphids in the field. III. Visual attraction. Annals of Applied Biology 49: 1 21. Kostal, V. & S. Finch, 1994. Influence of background on host plant selection and subsequent oviposition by the cabbage root fly (Delia radicum). Entomologia Experimentalis et Applicata 70: 153 163. Kostal, V. & S. Finch, 1996. Preference of the cabbage root fly, Delia radicum (L), for coloured traps: influence of sex and physiological status of the flies, trap background and experimental design. Physiological Entomology 21: 123 130. Moericke, V. 1952. Farben als Landereize für geflügelten Blattläusen (Aphidoidea). Zeitschrift für Naturforschung 7: 304 324. O Donnell, M. S. & T. H. Coaker, 1975. Potential of intracrop diversity for the control of brassica pests. Proceedings of the 8th British Insecticide and Fungicide Conference 1975, pp. 101 107. Perrin, R. M., 1977. Pest management in multiple cropping systems. Agro-ecosystems 3: 93 118. Perrin, R. M. & M. L. Phillips, 1978. Some effects of mixed cropping on the population dynamics of insect pests. Entomologia Experimentalis et Applicata 24: 385 393. Pimentel, D., 1961. The influence of plant spatial patterns on insect populations. Annals of the Entomological Society of America 54: 61 69. Prokopy, R. J. & R. D. Owens, 1983. Visual detection of plants by herbivorous insects. Annual Review of Entomology 28: 337 364. Prokopy, R. J., R. H. Collier & S. Finch, 1983. Visual detection of host plants by cabbage root flies. Entomologia Experimentalis et Applicata 34: 85 89. Richards, O. W., 1940. The biology of the small white butterfly (Pieris rapae) with special reference to the factors controlling its abundance. Journal of Animal Ecology 9: 243 288. Risch, S. J., D. Andow & M. A. Altieri, 1983. Agroecosystem diversity and pest control: Data, tentative conclusions and new research directions. Environmental Entomology 12: 625 629. Rojas J. C. & T. D. Wyatt, 1999. Role of visual cues and interaction with host odour during the host finding behaviour of the cabbage moth. Entomologia Experimentalis et Applicata 91: 59 65. Root, R. B., 1973. Organization of a plant-arthropod association in simple and diverse habitats: the fauna of collards (Brassica oleracea). Ecological Monograph 43: 95 124. Roitberg, B. D., I. C. Robertson & J. G. A. Tyerman, 1999. Vive la différence: a functional theory for insect herbivores. Entomologia Experimentalis et Applicata 91: 187 194. Rovira, A. D., 1969. Plant root exudates. Botanical Review 35: 35 39. Ryan, J., M. F. Ryan & F. McNaeidhe, 1980. The effect of interrow plant cover on populations of cabbage rootfly, Delia brassicae (Wied.). Journal of Applied Ecology 17: 31 40. Schoonhoven, L. M., 1968. Chemosensory bases of host plant selection. Annual Review of Entomology 13: 115 136. Simpson, S., A. J. Mordue (Luntz) & J. Hardie (guest eds.), 1999. Proceedings of the Tenth International Symposium on Insect- Plant Relationships. Entomologia Experimentalis et Applicata 91: 265 pp. Southwood, T. R. E., 1978. Ecological Methods with Particular Reference to the Study of Insect Populations. Chapman & Hall, London, 524 pp. Smith, J. G., 1969. Some effects of crop background on the populations of aphids and their natural enemies on Brussels sprouts. Annals of Applied Biology 63: 326 330. Smith, J. G., 1976. Influence of crop backgrounds on aphids and other phytophagous insects on Brussels sprouts. Annals of Applied Biology 83: 1 13. Tahvanainen, J. O. & R. B. Root, 1972. The influence of vegetational diversity on the population ecology of a specialized herbivore, Phyllotreta crucifera (Coleoptera: Chrysomelidae). Oecologia 10: 321 346. Theunissen, J., 1994. Effects of intercropping on pest populations in vegetable crops. IOBC/WPRS Bulletin 17 (8): 153 158. Theunissen, J. & H. den Ouden, 1980. Effects of intercropping with Spergula arvensis on pests of Brussels sprouts. Entomologia Experimentalis et Applicata 27: 260 268. Theunissen, J. & G. Schelling, 1996. Undersowing crops of white cabbage with strawberry clover and spurrey. IOBC/WPRS Bulletin 19 (11): 128 135. Theunissen, J., C. J. H. Booij & L. A. P. Lotz, 1995. Effects of intercropping white cabbage with clovers on pest infestation and yield. Entomologia Experimentalis et Applicata 74: 7 16. Theunissen, J., C. J. H. Booij, G. Schelling & J. Noorlander, 1992. Intercropping white cabbage with clover. IOBC/WPRS Bulletin XV/4: 104 114. Thiery, D. & J. H. Visser, 1986. Masking of host plant odour in the olfactory orientation of the Colorado potato beetle. Entomologia Experimentalis et Applicata 41: 165 172. Thorsteinson, A. J., 1960. Host selection in phytophagous insects. Annual Review of Entomology 5: 193 218. Tukahirwa, E. M. & T. H. Coaker, 1982. Effects of mixed cropping on some insect pests of brassicas; reduced Brevicoryne brassicae infestations and influences of epigeal predators and the disturbance of oviposition behaviour in Delia brassicae. Entomologia Experimentalis et Applicata 32: 129 140. Uvah, I. I. I. & T. H. Coaker, 1984. Effect of mixed cropping on some insect pests of carrots and onions. Entomologia Experimentalis et Applicata 36: 159 167. Vandermeer, J., 1989. The Ecology of Intercropping. Cambridge University Press, Cambridge, 237 pp. Verschaffelt, E., 1910. The cause determining the selection of food in some herbivorous insects. Proceedings of the Koninklijke Nederlandse Akademie van Wetenschappen 13: 536 542. Wilson, E. O., 1970. Chemical communication within animal species. In: E. Sondheimer & J. B. Simeone (eds). Chemical Ecology. Academic Press, New York and London, pp. 133 155. Wright, R. H., 1958. The olfactory guidance of flying insects. Canadian Entomologist 90: 81 89.